A method and system for analyzing the full life cycle of hidden dangers

Abstract

The present invention relates to the technical field of intrinsic safety and intelligent operation and inspection of power transmission lines, and in particular to a method and system for analyzing the full life cycle of hidden dangers in time series. The method comprises: firstly, acquiring multi-period acquired images by acquiring regional information of power transmission channels and transit satellite information of corresponding regions, calibrating the multi-period acquired images, acquiring multi-period registered images, identifying transmission equipment features on the multi-period registered images, extracting hidden danger features according to the characteristics of the power transmission equipment, acquiring hidden danger features of transmission channels, analyzing hidden danger changes on the multi-period registered images according to the hidden danger features of the transmission channels, predicting future hidden dangers according to meteorological data, and maintaining power transmission equipment according to the hidden danger prediction results. The present invention effectively solves the deficiencies of traditional hidden danger prediction methods in multi-source data integration and dynamic updating, effectively improves the timeliness and accuracy of hidden danger prediction, and enhances the safety of power transmission systems by comprehensively utilizing multi-period satellite images and real-time meteorological data.

Description

Technical Field

[0001] The present invention relates to the technical field of intrinsic safety and intelligent operation and inspection of power transmission lines, and in particular to a method and system for analyzing the full life cycle of hidden danger timing. Background Art

[0002] In the field of modern power transmission systems, with the rapid development of smart grid technology, real-time prediction and precise management of hidden dangers in transmission channels are becoming increasingly important. However, although current technology has made certain progress in the collection and preliminary processing of hidden danger data, how to comprehensively utilize multi-period satellite images and real-time meteorological data to conduct full life cycle analysis and accurate prediction of hidden dangers still faces major challenges.

[0003] Traditional methods are often unable to effectively integrate multi-dimensional data and lack a dynamic update mechanism, resulting in the inability to timely reflect the timeliness changes of meteorological factors and the dynamic registration of multi-period remote sensing images in a highly dynamic environment. Existing technologies are unable to reflect the changing trends of hidden dangers in real time, which in turn affects the safe operation and maintenance decisions of the transmission system.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art known to those skilled in the art. Summary of the invention

[0005] The present invention provides a method and system for analyzing the full life cycle of hidden dangers in a time sequence, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for analyzing the full life cycle of hidden dangers in a time series, the method comprising:

[0008] Collecting transmission channel area information and transit satellite information of the corresponding area, and obtaining multiple phases of collected images according to the transmission channel area information and the transit satellite information;

[0009] Performing image calibration on the multi-phase collected images to obtain multi-phase registered images;

[0010] Identify the power transmission equipment according to the multi-phase registered images to obtain the features of the power transmission equipment, extract hidden danger features from the multi-phase registered images according to the features of the power transmission equipment, and obtain hidden danger features of the power transmission channel;

[0011] Performing hidden danger change analysis on the multi-phase registered images according to the hidden danger characteristics of the power transmission channel to obtain hidden danger change trends;

[0012] The meteorological data corresponding to the multi-period registered images are collected, future hidden dangers are predicted according to the hidden danger change trend and the meteorological data, hidden danger prediction results are obtained, and the power transmission equipment is maintained based on the hidden danger prediction results.

[0013] Furthermore, image calibration is performed on the multi-phase collected images to obtain multi-phase registered images, including:

[0014] Analyzing the cloud coverage of the images collected in multiple periods to obtain image cloud information;

[0015] The multi-period collected images are screened according to the image cloud information to determine whether the cloud information of each period affects the hidden danger feature identification of the corresponding image collected in the current period, and if so, the images are eliminated to obtain multi-period screened images;

[0016] Sorting the multi-stage screening images according to a time series, marking any one of the multi-stage screening images as a first image, and marking the remaining multi-stage screening images as a second image set;

[0017] Matching the geometric position of the first image with the geographical position one by one, and eliminating the vertical and horizontal errors of the corresponding geographical position in the first image, to obtain a first pre-corrected image;

[0018] orthorectifying the first pre-corrected image to obtain a first orthorectified image, using the first orthorectified image as a base map, and orthorectifying the second image set respectively; obtaining a second orthorectified image set;

[0019] Taking the first orthorectified image as the origin, the second orthorectified image set is respectively registered with the first orthorectified image to obtain a multi-phase registered image.

[0020] Further, the cloud coverage rate of the multi-period collected images is analyzed to obtain the image cloud amount information, including:

[0021] collecting cloud layer reflection characteristics, and setting a cloud layer reflection threshold according to the cloud layer reflection characteristics;

[0022] A multi-band spectrum is used to perform spectral analysis on the multi-phase collected images to obtain the reflectivity of the multi-phase collected images, the reflectivity is compared with the cloud layer reflectivity threshold, the part with the reflectivity higher than the cloud layer reflectivity threshold is calibrated as the cloud layer, and the cloud measurement coverage rate is obtained.

[0023] Further, according to the hidden danger characteristics of the power transmission channel, hidden danger change analysis is performed on the multi-phase registered images to obtain hidden danger change trends, including:

[0024] Clustering and superimposing the multi-period registered images, and performing feature recognition on the superimposed images of the multi-period registered images according to the hidden danger characteristics of the power transmission channel to obtain a hidden danger superimposed image;

[0025] Collecting historical hidden danger information, and setting hidden variable scalar according to the historical hidden danger information;

[0026] Perform hidden danger change detection on the hidden danger superimposed image to obtain a number of hidden variables, compare the hidden variables with the hidden change scalar respectively, if the hidden variable is greater than the hidden change scalar, the corresponding pixel point is marked as 1, indicating an obvious change domain, if it is less than or equal to the hidden change scalar, it is marked as 0, indicating a stable domain, and obtain a hidden danger analysis result, wherein the hidden variable represents a quantized value obtained by the hidden danger change detection;

[0027] Based on the hidden danger analysis results, an image segmentation method is used to generate a latent change map, and the changing trend of hidden dangers in the transmission channel is obtained.

[0028] Furthermore, meteorological data corresponding to the multi-period registered images are collected, and future hidden dangers are predicted according to the hidden danger change trend and the meteorological data, including:

[0029] Extracting a number of meteorological factors according to the meteorological data, and assigning influencing factor weights to the number of meteorological factors according to the changing trend of hidden dangers in the power transmission channel;

[0030] Acquire future meteorological data according to meteorological satellites, perform weighted calculations according to the future meteorological data and the corresponding weights of the influencing factors to obtain weighted influence values, and aggregate several weighted influence values ​​to obtain a total weighted influence value;

[0031] Using the time series analysis method combined with the total weighted impact value to obtain the future hidden danger prediction value;

[0032] Collect historical hidden danger information to set a hidden danger threshold, compare the future hidden danger prediction value with the hidden danger threshold, and obtain a prediction result.

[0033] Furthermore, according to the changing trend of the hidden dangers of the power transmission channel, the weights of influencing factors are allocated to the several meteorological factors, including:

[0034] Extracting sampling data of a plurality of meteorological factors respectively according to the meteorological data, and performing standard normalization on the sampling data to obtain meteorological standard parameters;

[0035] Based on the time series, the distribution probabilities of several meteorological standard parameters at different time nodes are calculated respectively, and the uncertainty of the distribution probabilities is calculated to obtain the uncertainty of several meteorological factors, and the uncertainty is combined with the distribution probability of the corresponding meteorological factors to obtain the contribution of the corresponding meteorological factors to the trend of hidden danger changes in the power transmission channel, and obtain the disturbance measurement, which is used to quantify the contribution of meteorological factors to the hidden danger trend;

[0036] The degrees of certainty of the meteorological factors are obtained according to the disturbance measurement, and the influence factor weights of the corresponding factors are obtained by performing a proportional operation between the degree of certainty of any meteorological factor and the degrees of certainty of all meteorological factors.

[0037] Further, the certainties of the several meteorological factors are obtained according to the disturbance metric, and the influence factor weights of the corresponding factors are obtained by performing a proportional operation between the certainty of any one of the meteorological factors and the certainty of all the meteorological factors, including:

[0038] The calculation formula of the disturbance measure is as follows:

[0039] ;

[0040] in, for The perturbation measure of the meteorological factor, Indicates the Meteorological factors in The distribution probability at the time node, Indicates the Uncertainty of meteorological factors;

[0041] The impact factor weight calculation formula is as follows:

[0042] ;

[0043] in, is the impact factor weight, represents the disturbance measure.

[0044] Furthermore, the time series analysis method is combined with the total weighted impact value to obtain the future hidden danger prediction value, including:

[0045] Collecting historical meteorological information corresponding to the historical hidden danger information, sorting the historical hidden danger information and the historical meteorological information in time series to obtain a hidden danger time series set;

[0046] Constructing a time series model, dividing the hidden danger time series set into a hidden danger training set and a hidden danger test set, and using the hidden danger training set and the hidden danger test set to train and verify the time series model respectively;

[0047] The total weighted impact value is introduced as an exogenous variable into the time series model using data analysis software, and the time series model is fitted. Future hidden dangers are predicted based on the fitted time series model to obtain future hidden danger prediction values.

[0048] Furthermore, the power transmission equipment is identified according to the multi-phase registered images to obtain the features of the power transmission equipment, and the hidden danger features are extracted from the multi-phase registered images according to the features of the power transmission equipment to obtain the hidden danger features of the power transmission channel, including:

[0049] Collecting an image of the power transmission device according to the power transmission device, and acquiring characteristic information of the power transmission device based on the image of the power transmission device;

[0050] Constructing a comprehensive identification model, and performing deep learning on the characteristics of the power transmission equipment and historical hidden danger information based on the comprehensive identification model to obtain deep learning results;

[0051] The power transmission equipment is marked and removed from the multi-phase registered images according to the deep learning results to obtain hidden danger features of the power transmission channel.

[0052] A hidden danger time sequence full life cycle analysis system, the system comprising:

[0053] A multi-phase image acquisition module collects the transmission channel area information and the transit satellite information of the corresponding area, and obtains multi-phase acquisition images according to the transmission channel area information and the transit satellite information;

[0054] An image registration module performs image calibration on the multi-phase collected images to obtain multi-phase registered images;

[0055] A hidden danger feature extraction module is used to identify the power transmission equipment according to the multi-phase registered images to obtain the features of the power transmission equipment, and to extract the hidden danger features from the multi-phase registered images according to the features of the power transmission equipment to obtain the hidden danger features of the power transmission channel;

[0056] A hidden danger change trend analysis module is used to analyze hidden danger changes on the multi-phase registered images according to the hidden danger characteristics of the power transmission channel to obtain hidden danger change trends;

[0057] The future hidden danger prediction module collects the meteorological data corresponding to the multi-period registered images, predicts future hidden dangers according to the hidden danger change trend and the meteorological data, obtains the hidden danger prediction results, and maintains the power transmission equipment based on the hidden danger prediction results.

[0058] The technical solution of the present invention can achieve the following technical effects:

[0059] It effectively solves the shortcomings of traditional hidden danger prediction methods in multi-source data integration and dynamic updating. By comprehensively utilizing multi-period satellite images and real-time meteorological data, it effectively improves the timeliness and accuracy of hidden danger prediction and enhances the safety of the transmission system.

[0060] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 It is a flow chart of the hidden danger time series full life cycle analysis method;

[0063] Figure 2 Obtain a flow chart for the trend of hidden danger changes;

[0064] Figure 3 Obtain process flow diagram for hidden danger prediction;

[0065] Figure 4 This is a structural diagram for assigning impact factor weights. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0068] Embodiment 1;

[0069] like Figure 1 As shown, the present application provides a method for analyzing the full life cycle of hidden dangers in a time sequence, the method comprising:

[0070] S10: Collecting the regional information of the power transmission channel and the transit satellite information of the corresponding area, and obtaining multiple phases of collected images according to the regional information of the power transmission channel and the transit satellite information;

[0071] S20: performing image calibration on the multi-phase collected images to obtain multi-phase registered images;

[0072] S30: identifying the power transmission equipment according to the multi-phase registered images, obtaining the features of the power transmission equipment, extracting the hidden danger features from the multi-phase registered images according to the features of the power transmission equipment, and obtaining the hidden danger features of the power transmission channel;

[0073] S40: Analyze the hidden danger changes of the multi-period registered images according to the hidden danger characteristics of the power transmission channel to obtain the hidden danger change trend;

[0074] S50: Collect meteorological data corresponding to the multi-period registered images, predict future hidden dangers according to the hidden danger change trend and meteorological data, obtain hidden danger prediction results, and maintain the transmission equipment based on the hidden danger prediction results.

[0075] Specifically, remote sensing data, GIS systems and other technologies can be used to collect information about the transmission channel area, including terrain, landforms, vegetation coverage, transportation networks, etc., and then satellite sensors such as high-resolution satellites or remote sensing satellites can be used to obtain transit satellite images. The collected images are then calibrated, including geometric correction, color balance and geometric position correction. A registration algorithm can be used to align and fuse all images to obtain multi-period registered images. Then, automated image recognition technology, such as convolutional neural networks, is used to identify transmission equipment. The identified targets include transmission towers, substations, and power transformers. Stations, wires and other equipment, extract the characteristics of transmission equipment, identify the transmission equipment from multiple collected images according to the characteristics of transmission equipment, extract the hidden danger characteristics according to the identified transmission equipment, apply image difference analysis to identify the hidden danger area, and use change detection algorithms such as differential image method to analyze the hidden danger changes in multiple images to determine the development trend of hidden dangers, and then collect meteorological data corresponding to the multi-phase registered images, including sunshine duration, precipitation, temperature, wind speed, etc. Based on the hidden danger change trend and meteorological data, machine learning algorithms can be used to predict future hidden dangers and perform maintenance based on the prediction results.

[0076] The technical solution of the present invention effectively solves the shortcomings of traditional hidden danger prediction methods in multi-source data integration and dynamic updating. By comprehensively utilizing multi-period satellite images and real-time meteorological data, the timeliness and accuracy of hidden danger prediction are effectively improved, and the safety of the power transmission system is enhanced.

[0077] Specifically, image calibration is performed on the multi-phase collected images to obtain multi-phase registered images, including:

[0078] Analyze the cloud coverage rate of images collected in multiple periods to obtain the image cloud information;

[0079] Screen the multi-period collected images according to the image cloud information, and judge whether the cloud information of each period affects the hidden danger feature identification of the corresponding image collected in that period. If so, remove the images to obtain multi-period screened images;

[0080] sorting the multi-stage screening images according to the time series, marking any multi-stage screening image as the first image, and marking the remaining multi-stage screening images as the second image set;

[0081] Matching the geometric position of the first image with the geographical position one by one, and eliminating the vertical and horizontal errors of the corresponding geographical position in the first image, to obtain a first pre-corrected image;

[0082] Orthorectifying the first pre-corrected image to obtain a first orthorectified image, using the first orthorectified image as a base map, and orthorectifying the second image set respectively; obtaining a second orthorectified image set;

[0083] Taking the first orthorectified image as the origin, the second orthorectified image set is respectively registered with the first orthorectified image to obtain multi-phase registered images.

[0084] As a preferred embodiment of the above embodiment, firstly, the cloud coverage rate of the multi-period collected images is analyzed to obtain the image cloud information. Based on the obtained cloud information, the cloud coverage of each period of images is analyzed. If the cloud coverage exceeds the set threshold, it indicates that the cloud layer in the image has a large impact and may interfere with the identification of hidden danger features. The images that are greatly affected by the cloud layer are eliminated to obtain multi-period screened images. Then, the images that have passed the cloud coverage screening are sorted according to the shooting time. From the sorted multi-period screened images, any image is marked as the first image, and the remaining images are marked as the second image set. A geographic information system or high-precision positioning data can be used to correspond the geometric position of the first image to the actual geographic coordinates one by one. By analyzing the coordinate system of the image, the geographic location in the image is eliminated. Errors can be corrected by using mathematical models and ground control points to perform orthorectification on the first pre-corrected image. This process eliminates terrain and camera distortion in the image so that the image is accurately aligned with the geographic data on the map coordinates. During the correction process, the image will be geometrically corrected according to factors such as the earth's curvature, shooting angle, and camera internal parameters. After the orthorectification is completed, the image obtained is the first orthorectified image. The first orthorectified image is used as the base map. According to the same orthorectification method, each image in the second image set is corrected to obtain the second orthorectified image set. For example, an image registration algorithm is used, and the first orthorectified image is used as a reference. Each image in the second orthorectified image set is registered with it. The registered images are merged into a set of multi-period registered images.

[0085] Furthermore, the cloud coverage rate of the multi-period collected images is analyzed to obtain the image cloud information, including:

[0086] Collect cloud reflection characteristics and set cloud reflection thresholds based on the cloud reflection characteristics;

[0087] Multi-band spectrum is used to perform spectral analysis on multi-period collected images to obtain the reflectivity of the multi-period collected images. The reflectivity is compared with the cloud reflection threshold, and the part with reflectivity higher than the cloud reflection threshold is calibrated as cloud layer to obtain the cloud measurement coverage.

[0088] In the embodiment itself, firstly, the spectral reflectance characteristics of different types of clouds, such as cumulus clouds and rain clouds, can be recorded through real ground observation data or meteorological satellite data, and the average reflectance of clouds in each spectral band can be statistically analyzed to determine a cloud reflection threshold. Multi-band spectral analysis technology is applied to the collected multi-period images, especially in the infrared and near-infrared bands, which are particularly sensitive to the identification of clouds. The reflectance of each pixel point of each image is compared with the previously set cloud reflection threshold. If the reflectance of a pixel point is higher than the cloud reflection threshold, the pixel point is calibrated as a cloud layer. The number of pixels calibrated as clouds in each image is counted, and compared with the total number of pixels, the proportion of cloud coverage is calculated. This proportion is expressed as the cloud measurement coverage rate, that is, the proportion of the area occupied by clouds in the image, thereby obtaining the cloud measurement coverage rate.

[0089] Further, if Figure 2 As shown in the figure, according to the hidden danger characteristics of the transmission channel, the hidden danger change analysis is performed on the multi-period registered images to obtain the hidden danger change trend, including:

[0090] S41: clustering and superimposing the multi-period registered images, and performing feature recognition on the images superimposed by the multi-period registered images according to the hidden danger characteristics of the power transmission channel to obtain a hidden danger superimposed image;

[0091] S42: Collect historical hidden danger information, and set hidden variable scalar according to the historical hidden danger information;

[0092] S43: Perform hidden danger change detection on the hidden danger superimposed image to obtain a number of hidden variables, compare the hidden variables with the hidden change scalar respectively, if the hidden variable is greater than the hidden change scalar, the corresponding pixel point is marked as 1, indicating an obvious change domain, if it is less than or equal to the hidden change scalar, it is marked as 0, indicating a stable domain, and obtain the hidden danger analysis result, the hidden variable represents the quantitative value obtained by the hidden danger change detection;

[0093] S44: Based on the hidden danger analysis results, the image segmentation method is used to generate a latent change map and obtain the change trend of the hidden dangers in the transmission channel.

[0094] Specifically, firstly, the registered images at multiple time points are clustered and superimposed, and the hidden danger features in each period of images are classified by using, for example, K-means clustering, and similar feature areas are grouped into one category. Then, a pixel-based clustering method is used to identify the distribution of hidden dangers in different areas, and a new synthetic image is generated. The image shows the change characteristics of the hidden danger area in different time periods. Feature recognition is performed on the clustered and superimposed images, especially focusing on the change characteristics of the hidden danger area, such as the increase, decrease, expansion or disappearance of the hidden danger. Then, based on the historical hidden danger information, such as the change amplitude and occurrence frequency of the hidden danger, the latent change scalar is set, and the hidden danger superimposed image is subjected to change detection, and the change of the hidden danger area at different time points is analyzed by image difference analysis, such as pixel difference method and image The registration difference method, etc., calculates the change of the hidden danger area in different time periods and quantifies it to obtain several latent variables. These latent variables represent the change of the hidden danger area at each time point, such as the change of the hidden danger area and the speed of hidden danger expansion. The obtained latent variables are compared with the preset latent change scalar one by one. If the latent variable is greater than the latent change scalar, it means that the hidden danger change in this area is significant and it is marked as the obvious change domain. If the latent variable is less than or equal to the latent change scalar, it means that the hidden danger change in this area is not obvious and it is marked as the stable domain. Then, according to the comparison result, each pixel point is marked, the pixel points in the obvious change domain are marked as 1, and the pixel points in the stable domain are marked as 0. Finally, the obvious change domain and the stable domain are segmented by methods such as K-means clustering to generate a latent change map, so as to obtain the change trend of the hidden dangers in the transmission channel.

[0095] Furthermore, if Figure 3 As shown in the figure, the meteorological data corresponding to the multi-period registered images are collected, and the future hidden dangers are predicted according to the hidden danger change trend and meteorological data, including:

[0096] S51: extracting several meteorological factors according to meteorological data, and assigning influencing factor weights to the several meteorological factors according to the changing trend of hidden dangers in the power transmission channel;

[0097] S52: acquiring future meteorological data according to meteorological satellites, performing weighted calculations according to the future meteorological data and corresponding influencing factor weights to obtain weighted influencing values, and summarizing several weighted influencing values ​​to obtain a total weighted influencing value;

[0098] S53: Use time series analysis combined with total weighted impact value to obtain future hidden danger prediction value;

[0099] S54: Collect historical hidden danger information to set a hidden danger threshold, compare the future hidden danger prediction value with the hidden danger threshold, and obtain a prediction result.

[0100] As a preferred embodiment of the above, firstly, corresponding real-time meteorological data including sunshine duration, precipitation, temperature, wind speed, humidity, etc. are obtained from meteorological satellites. According to the obtained trend of hidden danger changes in the power transmission channel, the main factors affecting the hidden danger changes are identified, and the weight of the influencing factor is assigned to each meteorological factor according to the degree of influence. The meteorological factors are weighted using statistical methods or machine learning models such as regression analysis. According to meteorological satellite data or other meteorological model prediction tools, future meteorological data such as precipitation, temperature, wind speed, etc. in the future are obtained. The future meteorological data is combined with the weight of the influencing factor, and weighted calculation is performed on each meteorological factor according to its weight. Specifically, the value of each meteorological factor Multiply it by its corresponding weight to get the weighted impact value, and then summarize all the weighted impact values ​​to get the total weighted impact value, that is, the overall impact of all meteorological factors on future hidden dangers. Subsequently, time series analysis methods, such as the ARIMA model, are used in combination with the total weighted impact value to analyze the historical change trend and future prediction value of the hidden dangers. Based on the fitting results of the time series model, future hidden dangers are predicted in combination with the total weighted impact value to obtain the hidden danger prediction value within a certain period in the future. The hidden danger threshold is set according to the historical hidden danger data, and the predicted future hidden danger value is compared with the set hidden danger threshold to determine whether the hidden danger exceeds the preset threshold. If the predicted value exceeds the threshold, it means that there is a large hidden danger and early warning and equipment maintenance are required.

[0101] Further, if Figure 4 As shown in the figure, according to the changing trend of hidden dangers in the transmission channel, several meteorological factors are assigned influencing factor weights, including:

[0102] According to the meteorological data, the sampling data of several meteorological factors are extracted respectively, and the sampling data are normalized to obtain the meteorological standard parameters;

[0103] Based on the time series, the distribution probabilities of several meteorological standard parameters at different time nodes are calculated respectively, and the distribution probabilities are calculated with uncertainty to obtain the uncertainties of several meteorological factors. The uncertainties are combined with the distribution probabilities of the corresponding meteorological factors to obtain the contribution of the corresponding meteorological factors to the trend of hidden danger changes in the power transmission channel, and the disturbance measurement is obtained. The disturbance measurement is used to quantify the contribution of meteorological factors to the trend of hidden dangers;

[0104] The certainties of several meteorological factors are obtained according to the disturbance measurement, and the influencing factor weights of the corresponding factors are obtained by performing a proportional operation between the certainty of any meteorological factor and the certainty of all meteorological factors.

[0105] In this embodiment, each meteorological factor is standardized and normalized according to the collected meteorological data, usually by minimum-maximum normalization or Z-score normalization. After the normalization is completed, the obtained meteorological data is converted into meteorological standard parameters, and then for each meteorological standard parameter at a non-stop time node, its distribution probability is calculated. By analyzing the distribution frequency of meteorological data at different time nodes, the distribution probability of each meteorological factor at each time point is obtained. Based on the distribution probability, the entropy calculation of information theory can be used, combined with the distribution probability of the meteorological factor and the calculated uncertainty of the meteorological factor, so as to quantify the contribution of the meteorological factor to the hidden danger change and obtain the disturbance measure. The disturbance measure represents the influence of the meteorological factor on the hidden danger trend change. The certainty of each meteorological factor is calculated through the value of the disturbance measure, that is, the decisive role of the meteorological factor in the prediction of hidden danger changes. The factor with higher certainty contributes more to the hidden danger prediction. The weight of the influencing factor of each meteorological factor is calculated by performing proportional operation on the certainty of each meteorological factor and the certainty of all meteorological factors.

[0106] Specifically, the certainties of several meteorological factors are obtained according to the disturbance measurement, and the influencing factor weights of the corresponding factors are obtained by performing a proportional operation between the certainty of any meteorological factor and the certainty of all meteorological factors, including:

[0107] The perturbation measure is calculated as follows:

[0108] ;

[0109] in, for Perturbation measure of meteorological factors, express Meteorological factors in The distribution probability at the time node, express Uncertainty of meteorological factors;

[0110] The impact factor weight calculation formula is as follows:

[0111] ;

[0112] in, is the impact factor weight, represents the perturbation measure.

[0113] As a preferred embodiment of the above embodiment, the above disturbance measurement formula is used to calculate the uncertainty of each meteorological factor at different time nodes, that is, to calculate the potential impact of the meteorological factor on the change of hidden dangers. Meteorological factors with high disturbance measurements indicate that they have a greater impact on the trend of hidden danger changes and usually require higher weights. Then, the disturbance measurement of each meteorological factor is performed to obtain its corresponding degree of certainty. Meteorological factors with high degrees of certainty indicate that their impact on the change of hidden dangers is relatively stable and clear, and usually occupy a larger proportion in the prediction. According to the degree of certainty of each meteorological factor, the influence factor weight of the factor is calculated through proportional operation, and the influence factor weight reflects the contribution of each meteorological factor to the change of hidden dangers.

[0114] Furthermore, the time series analysis method is combined with the total weighted impact value to obtain the future hidden danger prediction value, including:

[0115] According to the historical meteorological information corresponding to the historical hidden danger information collection, the historical hidden danger information and the historical meteorological information are sorted in time series to obtain a hidden danger time series set;

[0116] Construct a time series model, divide the hidden danger time series set into a hidden danger training set and a hidden danger test set, and use the hidden danger training set and the hidden danger test set to train and verify the time series model respectively;

[0117] The total weighted impact value is introduced into the time series model as an exogenous variable using data analysis software, and the time series model is fitted. Future hidden dangers are predicted based on the fitted time series model to obtain future hidden danger prediction values.

[0118] In this embodiment, by collecting historical meteorological information corresponding to historical hidden danger information and arranging it in chronological order, a hidden danger time series set is constructed, which includes hidden danger information and corresponding meteorological data at each time node. As the input data set of the time series model, ARIMA and other models can be selected to construct the time series model, and the acquired hidden danger time series set is divided into a training set and a test set. Usually, 80% of the data is used for training and 20% of the data is used for verification. Then, time series modeling is performed on the training set, and the model is learned using the data of the training set to fit the relationship between the hidden danger data and the meteorological factors. The trained model is verified using the test set data to evaluate the prediction ability of the model on unseen data. In the model fitting process, the obtained total weighted influence value is used as an exogenous variable, that is, an additional variable in the model, and an autoregressive model is used to introduce the exogenous variable into the time series model. Finally, the fitted time series model is used to predict the hidden danger change trend at future time nodes based on the historical hidden danger information, meteorological data and the total weighted influence value.

[0119] The power transmission equipment is identified based on the multi-phase registered images to obtain the features of the power transmission equipment. The hidden danger features of the multi-phase registered images are extracted based on the features of the power transmission equipment to obtain the hidden danger features of the power transmission channel, including:

[0120] Collecting images of power transmission equipment according to the power transmission equipment, and obtaining characteristic information of the power transmission equipment based on the images of the power transmission equipment;

[0121] Build a comprehensive identification model, conduct deep learning on the characteristics of power transmission equipment and historical hidden danger information based on the comprehensive identification model, and obtain deep learning results;

[0122] Based on the deep learning results, the power transmission equipment is marked and removed from the multi-phase registered images to obtain the hidden danger characteristics of the power transmission channel.

[0123] Specifically, multi-phase satellite remote sensing images or drone images are used to collect image data of power transmission equipment. The collected images can cover all equipment in the transmission channel, such as transmission towers, conductors, substations, etc. Based on the collected transmission equipment images, image processing technology is used to extract the geometric features of the equipment, including position, size, shape, direction, etc. Based on deep learning technology, a comprehensive transmission equipment recognition model is constructed, and the model is trained using historical hidden danger information and transmission equipment feature information to identify the characteristics of the equipment and possible hidden dangers of the equipment. According to the recognition results of the deep learning model, the transmission equipment in the multi-phase registered images is marked. Based on the marked equipment area, the difference image method, edge detection method and other methods can be used to extract possible hidden danger features, such as damage to the appearance of the equipment, changes in the environment around the equipment, signs of aging, etc., to finally obtain the hidden danger characteristics of the transmission channel.

[0124] Embodiment 2:

[0125] Based on the same inventive concept as the method for analyzing the full life cycle of a hidden danger sequence in the aforementioned embodiment, the present invention further provides a system for analyzing the full life cycle of a hidden danger sequence, the system comprising:

[0126] A multi-phase image acquisition module collects the regional information of the power transmission channel and the transit satellite information of the corresponding area, and obtains multi-phase acquisition images based on the regional information of the power transmission channel and the transit satellite information;

[0127] An image registration module performs image calibration on multi-phase collected images to obtain multi-phase registered images;

[0128] The hidden danger feature extraction module identifies the power transmission equipment based on the multi-phase registered images, obtains the features of the power transmission equipment, extracts the hidden danger features from the multi-phase registered images based on the features of the power transmission equipment, and obtains the hidden danger features of the power transmission channel;

[0129] The hidden danger change trend analysis module analyzes the hidden danger changes of multi-period registered images according to the hidden danger characteristics of the transmission channel to obtain the hidden danger change trend;

[0130] The future hidden danger prediction module collects meteorological data corresponding to multiple periods of registered images, predicts future hidden dangers according to the hidden danger change trend and meteorological data, obtains hidden danger prediction results, and maintains the transmission equipment based on the hidden danger prediction results.

[0131] The above-mentioned adjustment system in the present invention can effectively implement the hidden danger timing full life cycle analysis method, and the technical effects that can be achieved are as described in the above-mentioned embodiments, which will not be repeated here.

[0132] Although the present application has been described in conjunction with specific features and embodiments thereof, it is obvious that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application as defined therein, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A method for analyzing the full life cycle of hidden dangers, characterized in that: The method comprises: Collecting transmission channel area information and transit satellite information of the corresponding area, and obtaining multiple phases of collected images according to the transmission channel area information and the transit satellite information; Performing image calibration on the multi-phase collected images to obtain multi-phase registered images; The method for obtaining the multi-phase registered images includes: Analyzing the cloud coverage of the images collected in multiple periods to obtain image cloud information; The multi-period collected images are screened according to the image cloud information to determine whether the cloud information of each period affects the hidden danger feature identification of the corresponding image collected in the current period, and if so, the images are eliminated to obtain multi-period screened images; Sorting the multi-stage screening images according to a time series, marking any one of the multi-stage screening images as a first image, and marking the remaining multi-stage screening images as a second image set; Matching the geometric position of the first image with the geographical position one by one, and eliminating the vertical and horizontal errors of the corresponding geographical position in the first image, to obtain a first pre-corrected image; orthorectifying the first pre-corrected image to obtain a first orthorectified image, using the first orthorectified image as a base map, and orthorectifying the second image set respectively; obtaining a second orthorectified image set; Taking the first orthorectified image as the origin, registering the second orthorectified image set with the first orthorectified image respectively to obtain a multi-phase registered image; Identify the power transmission equipment according to the multi-phase registered images to obtain the features of the power transmission equipment, extract hidden danger features from the multi-phase registered images according to the features of the power transmission equipment, and obtain hidden danger features of the power transmission channel; Performing hidden danger change analysis on the multi-phase registered images according to the hidden danger characteristics of the power transmission channel to obtain hidden danger change trends; Collecting meteorological data corresponding to the multi-period registered images, predicting future hidden dangers according to the hidden danger change trend and the meteorological data, obtaining hidden danger prediction results, and maintaining the power transmission equipment based on the hidden danger prediction results; The future hidden danger prediction method comprises: Extracting a number of meteorological factors according to the meteorological data, and assigning influencing factor weights to the number of meteorological factors according to the changing trend of hidden dangers in the power transmission channel; Assigning the impact factor weights includes: Extracting sampling data of a plurality of meteorological factors respectively according to the meteorological data, and performing standard normalization on the sampling data to obtain meteorological standard parameters; Based on the time series, the distribution probabilities of several meteorological standard parameters at different time nodes are calculated respectively, and the uncertainty of the distribution probabilities is calculated to obtain the uncertainty of several meteorological factors, and the uncertainty is combined with the distribution probability of the corresponding meteorological factors to obtain the contribution of the corresponding meteorological factors to the trend of hidden danger changes in the power transmission channel, and obtain the disturbance measurement, which is used to quantify the contribution of meteorological factors to the hidden danger trend; Obtaining the degrees of certainty of the several meteorological factors according to the disturbance measurement, and obtaining the influence factor weight of the corresponding factor by performing a proportional operation on the degree of certainty of any meteorological factor and the degrees of certainty of all meteorological factors; The method for obtaining the influencing factor weight of the corresponding factor includes: The calculation formula of the disturbance measure is as follows: ； in, for The perturbation measure of the meteorological factor, Indicates the Meteorological factors in The distribution probability at the time node, Indicates the The uncertainty of meteorological factors, Indicates The meteorological factors; The impact factor weight calculation formula is as follows: ； in, is the impact factor weight, represents the perturbation measure, Indicates The meteorological factors, Indicates a total of The meteorological factors; Acquire future meteorological data according to meteorological satellites, perform weighted calculations according to the future meteorological data and the corresponding weights of the influencing factors to obtain weighted influence values, and aggregate several weighted influence values ​​to obtain a total weighted influence value; Using the time series analysis method combined with the total weighted impact value to obtain the future hidden danger prediction value; The method for obtaining the future hidden danger prediction value includes: Collecting historical meteorological information corresponding to the historical hidden danger information, sorting the historical hidden danger information and the historical meteorological information in time series to obtain a hidden danger time series set; Constructing a time series model, dividing the hidden danger time series set into a hidden danger training set and a hidden danger test set, and using the hidden danger training set and the hidden danger test set to train and verify the time series model respectively; Using data analysis software to introduce the total weighted impact value as an exogenous variable into the time series model, and fitting the time series model, predicting future hidden dangers according to the fitted time series model to obtain future hidden danger prediction values; Collect historical hidden danger information to set a hidden danger threshold, compare the future hidden danger prediction value with the hidden danger threshold, and obtain a prediction result.

2. The method for analyzing the hidden danger time series life cycle according to claim 1 is characterized in that: Performing cloud coverage analysis on the multiple collected images to obtain image cloud information includes: collecting cloud layer reflection characteristics, and setting a cloud layer reflection threshold according to the cloud layer reflection characteristics; A multi-band spectrum is used to perform spectral analysis on the multi-phase images to obtain the reflectivity of the multi-phase collected images, the reflectivity is compared with the cloud layer reflectivity threshold, the part with the reflectivity higher than the cloud layer reflectivity threshold is calibrated as the cloud layer, and the cloud measurement coverage is obtained.

3. The method for analyzing the full life cycle of hidden dangers according to claim 1 is characterized in that: According to the hidden danger characteristics of the power transmission channel, the hidden danger change analysis is performed on the multi-phase registered images to obtain the hidden danger change trend, including: Clustering and superimposing the multi-period registered images, and performing feature recognition on the superimposed images of the multi-period registered images according to the hidden danger characteristics of the power transmission channel to obtain a hidden danger superimposed image; Collecting historical hidden danger information, and setting hidden variable scalar according to the historical hidden danger information; Perform hidden danger change detection on the hidden danger superposition information to obtain a number of hidden variables, compare the hidden variables with the hidden change scalar respectively, if the hidden variable is greater than the hidden change scalar, the corresponding pixel point is marked as 1, indicating an obvious change domain, if it is less than or equal to the hidden change scalar, it is marked as 0, indicating a stable domain, and obtain a hidden danger analysis result, wherein the hidden variable represents a quantized value obtained by the hidden danger change detection; Based on the hidden danger analysis results, an image segmentation method is used to generate a latent change map, and the changing trend of hidden dangers in the transmission channel is obtained.

4. The method for analyzing the full life cycle of hidden dangers according to claim 1 is characterized in that: The power transmission equipment is identified according to the multi-phase registered images to obtain the features of the power transmission equipment, and the hidden danger features are extracted from the multi-phase registered images according to the features of the power transmission equipment to obtain the hidden danger features of the power transmission channel, including: Collecting an image of the power transmission device according to the power transmission device, and acquiring characteristic information of the power transmission device based on the image of the power transmission device; Constructing a comprehensive identification model, and performing deep learning on the characteristics of the power transmission equipment and historical hidden danger information based on the comprehensive identification model to obtain deep learning results; The power transmission equipment is marked and removed from the multi-phase registered images according to the deep learning results to obtain hidden danger features of the power transmission channel.

5. A hidden danger time series full life cycle analysis system, characterized in that: Using the hidden danger time series full life cycle analysis method as claimed in claim 1, the system comprises: A multi-phase image acquisition module collects the transmission channel area information and the transit satellite information of the corresponding area, and obtains multi-phase acquisition images according to the transmission channel area information and the transit satellite information; An image registration module performs image calibration on the multi-phase collected images to obtain multi-phase registered images; A hidden danger feature extraction module is used to identify the power transmission equipment according to the multi-phase registered images to obtain the features of the power transmission equipment, and to extract the hidden danger features from the multi-phase registered images according to the features of the power transmission equipment to obtain the hidden danger features of the power transmission channel; A hidden danger change trend analysis module performs hidden danger change analysis on the multi-phase registered images according to the hidden danger characteristics of the power transmission channel to obtain the hidden danger change trend; The future hidden danger prediction module collects the meteorological data corresponding to the multi-period registered images, predicts future hidden dangers according to the hidden danger change trend and the meteorological data, obtains the hidden danger prediction results, and maintains the power transmission equipment based on the hidden danger prediction results.

Images