Method and system for the analysis of monitoring data of electrcity distribution networks for the identification of the components present in said network and any anomalies thereof

Abstract

Method and system for the analysis of monitoring data of electricity distribution networks for the identification of the components present in said network and any anomalies thereof, the method providing the following steps: providing for the acquisition of monitoring data by means of at least one, preferably two or more techniques for acquiring said data; the at least one monitoring data acquisition technique providing for the interaction of one or more components present in said network and / or of one or more objects provided in the environment surrounding said components with electromagnetic radiation, in particular said at least one acquisition technique being of the photographic type or being a remote sensing technique of the so-called LIDAR (Laser Imaging Detection and Ranging) type, or consisting of satellite images; processing the monitoring data acquired with at least one of said acquisition techniques for the identification of said components present in said distribution network and / or of said one or more objects present in the environment surrounding said network, and determining from said anomaly monitoring data one or more of said components, wherein the monitoring data collected according to at least one of said acquisition techniques are processed by means of combinations of classification algorithms and progressive segmentation of the monitoring data.

Description

[0001] “METHOD AND SYSTEM FOR THE ANALYSIS OF MONITORING DATA OF ELECTRCITY DISTRIBUTION NETWORKS FOR THE INDENTIFICATION OF THE COMPONENTS PRESENT IN SAID NETWORK AND ANY ANOMALIES THEREOF”

[0002] FIELD OF THE INVENTION

[0003] The present invention concerns a method, and a system for implementing such method, for the analysis of monitoring data of electricity distribution networks for the identification of the components present in such network and any anomalies thereof

[0004] BACKGROUND OF THE INVENTION

[0005] Currently, the process of acquiring data is followed in a known maimer for different types of acquisition techniques. In the case of images, these are for example photographic images acquired from ground positions or from in-flight positions.

[0006] Similarly, for the acquisition of three-dimensional point clouds, by means of LIDAR technology or suchlike, the acquisition can take place by means of sensors operating on the ground or mounted on aircraft.

[0007] A huge amount of data is acquired for each distribution network, and in the case of photographic monitoring techniques the number of images is very high.

[0008] Currently, the analysis to identify the presence, in the various images, of a component provided in the distribution network and / or an object present in the environment surrounding the component and / or the network is performed manually by specialized teams of people, and the cost of this operation is very high, both in economic terms and also in terms of the speed at which the analysis is performed.

[0009] In addition, the analysis performed by human personnel is affected by evaluation errors that are caused by oversights or incorrect interpretations of the images, and which at the same time are also affected by the structural peculiarities of the networks and technical know-how of the personnel in different geographical regions.

[0010] Since the amount of data is very high, the use of generic type artificial intelligence systems also presents problems. Document US 2022 / 131375 Al describes a method for managing the power grid of an electric utility comprising a utility power line route. The method includes generating an infrastructure model comprising utility assets and image information; receiving power line imaging data collected on the route and including public utility asset imaging data; receiving geospatial topological model information; and generating a navigable simulated virtual environment model including an integrated visualization of infrastructure model image information with geospatial topological model image information and power line imaging data. A trained classification algorithm classifies the conditions of the utility assets on the route from virtual views.

[0011] Document US 2020 / 082168 Al describes various examples for the acquisition, processing and output generation of data to be used in the analysis of a physical asset or a collection of physical assets of interest.

[0012] There is therefore the need to perfect a method, and a system for implementing such method, for the analysis of monitoring data of electricity distribution networks for the identification of the components present in such network and any anomalies thereof, which can overcome at least one of the disadvantages of the state of the art.

[0013] In particular, the present invention aims to provide a method, and a system for implementing such method, for the analysis of monitoring data of electricity distribution networks for the identification of the components present in such network and any anomalies thereof, which can be largely automated, guaranteeing reasonable costs and processing times, as well as high performances in the precision and reliability of the results which are at least similar and possibly better than those achieved through analysis by human personnel.

[0014] The Applicant has devised, tested and embodied the present invention to overcome the shortcomings of the state of the art and to obtain these and other purposes and advantages.

[0015] SUMMARY OF THE INVENTION

[0016] The present invention is set forth and characterized in the independent claims, while the dependent claims describe other characteristics of the present invention or variants to the main inventive idea.

[0017] In accordance with the above purposes, some embodiments described here concern a method for the analysis of monitoring data of electricity distribution networks for the identification of the components present in such network and any anomalies thereof, the method providing the following steps: providing for the acquisition of monitoring data by means of at least one, preferably two or more techniques for acquiring such data; the at least one monitoring data acquisition technique providing for the interaction of one or more components present in the network and / or of one or more objects provided in an environment surrounding the components with electromagnetic radiation, in particular the at least one acquisition technique being of the photographic type or being a remote sensing technique of the so-called LIDAR (Laser Imaging Detection and Ranging) type, or consisting of satellite images; the alternative or combined execution of the acquisition of the monitoring data according to two or more of the acquisition techniques as above; the transformation of the monitoring data according to the at least two or more acquisition techniques into images and the processing of the images relating to the identification of the components present in the distribution network and / or of the one or more objects present in the environment surrounding the network and the determination, from the images, of any anomalies of one or more of the components.

[0018] The term anomalies of the components that are provided in a network is understood within the scope of the present invention in an extended sense and also comprises anomalous conditions in reference to objects present in the environment surrounding the distribution network, such as in particular geometric distances between one or more components and one or more objects present in the environment surrounding the distribution network and / or interference of any kind between one or more components present in the distribution network and one or more objects present in the surrounding environment.

[0019] According to a first embodiment, the invention provides a method for the analysis of monitoring data of electricity distribution networks for the identification of the components present in such network and any anomalies thereof, the method providing the following steps: providing for the acquisition of monitoring data by means of at least one, preferably two or more techniques for acquiring such data; the at least one monitoring data acquisition technique providing for the interaction of one or more components present in the network and / or of one or more objects provided in an environment surrounding the components with electromagnetic radiation, in particular the at least one acquisition technique being of the photographic type or being a remote sensing technique of the so-called LIDAR (Laser Imaging Detection and Ranging) type, or consisting of satellite images; the alternative or combined execution of the acquisition of the monitoring data according to two or more of the acquisition techniques as above; processing the monitoring data acquired with at least one of the acquisition techniques for the identification of the components present in the distribution network and / or of the one or more objects present in the environment surrounding the network and determining, from the anomaly monitoring data, one or more of the components, wherein the monitoring data acquired according to at least one of the acquisition techniques are processed by means of combinations of algorithms for the classification and progressive segmentation of the monitoring data.

[0020] According to one embodiment, the monitoring data consist of both photographic images, preferably in high definition, and also a 3D point cloud acquired by means of LIDAR technology, or suchlike, the monitoring data being processed by means of machine learning algorithms specifically trained, respectively, for the processing of photographic images and for the processing of 3D point clouds acquired by means of LIDAR technology, or suchlike.

[0021] In one embodiment, at least for the monitoring data consisting of photographic images, the method provides to subject the photographic images, or at least certain regions thereof, to an automatic evaluation of the quality of the image.

[0022] Since the quality of an image can also be a subjective matter, the method provides one or more of the following parameters as a quality parameter: signal to noise ratio; sharpness of contours; contrast; focus; definition.

[0023] It is possible to provide, alternatively or in combination, other parameters for measuring image quality which can be selected following specific conditions.

[0024] In particular, in one embodiment, the method provides to determine a threshold value corresponding to a minimum threshold value of acceptable quality, and to determine the same value for the image or images of the electricity distribution network to be processed, eliminating from the monitoring data to be processed those relating to the images whose image quality value is lower than that of the minimum quality threshold.

[0025] In one embodiment, when it is provided to use a photographic technique, preferably high definition, as a monitoring data acquisition technique, the present invention provides to carry out the following steps of processing the monitoring data consisting of the photographic images: a first object detection step in which, by means of a first computer vision algorithm model, regions of interest are identified in the photographic images in which one or more components of the distribution network are represented, respectively; subsequently, in a second processing step, the various regions of interest identified in the first step, optionally only said regions of interest, are subjected to processing by means of one or more models of additional computer vision algorithms specialized for the identification of each component of the distribution network which can potentially be provided therein; at the end of this step, a third processing step is performed in which the image is cropped, generating an image cutout for one or more of the components identified in the previous step, and the image cutouts are subjected to one or more additional models of computer vision algorithms specialized, respectively, for the identification of the anomalies of the components and / or of the materials which the components are made of and / or of optional further features of the components identified in the photographic images.

[0026] According to one variant, the one or more additional models of computer vision algorithms, in particular those applied for processing the image cutouts, consist of classification and / or semantic segmentation algorithms. Providing a plurality of models of computer vision algorithms, which are specialized, respectively, to identify the regions of interest in an image, i.e. the regions of the image that contain the target objects, in this case one or more components of the distribution network, to recognize a typology of objects in each of the zones of interest, and to identify anomalies and / or materials and / or other parameters relating to the components in cutouts of the image and of the regions of interest which contain the components, considerably reduces the amount of data to be subjected to processing, progressively restricting the data to be subjected to processing for the identification steps that require greater precision since they relate to essential information, such as the results of the analysis, i.e. type of component, functional state, material and other parameters of the components that are identifiers of a functional condition thereof.

[0027] According to a preferred embodiment, before the first processing step and / or preferably before the second and / or preferably before the third processing step, the regions of interest of the image and / or the cutouts of the regions of interest containing the image of one or more components of the distribution network are subjected to a further processing in order to verify image quality, wherein: quality parameters of the image are measured, in particular image definition and / or contrasts and / or sharpness and / or focus, or optionally other parameters, and the values are compared with pre-set threshold values which define a minimum image quality threshold, the image cutouts for which the result of the comparison reveals values of the parameters lower than the minimum image quality threshold being eliminated from the processing process of the step provided after the quality analysis.

[0028] According to other characteristics of the invention, as an alternative to or in combination with one or more of the previous characteristics, the method provides that a 3D point cloud acquisition technique, such as a LIDAR type technology, or suchlike, is used as the monitoring data acquisition technique.

[0029] In this case, the present invention provides to carry out the following succession of steps of processing the monitoring data consisting of the 3D point cloud: a first step of semantic segmentation of the 3D point cloud, each component being segmented with reference to its own geometry; a second step of identifying the components of the electricity distribution network and / or objects present in the surrounding environment, such as vegetation, artefacts, such as cabins, buildings and other supports such as poles, or suchlike, and of the cables present in the distribution network subjected to monitoring; the generation of a map of the electricity distribution network and / or of the surrounding environment; the calculation of anomalies relating to the distances between one or more components of the electricity distribution network and / or one or more objects of the surrounding environment and / or of interferences of other types between the components and the objects of the surrounding environment.

[0030] As regards the step of generating a map of the electricity distribution network and / or of the surrounding environment, it is possible to use an extraction method in which, following the processing relating to the semantic segmentation of the individual points of the point cloud and the identification of the components of the electricity network and / or of the surrounding environment, the points are projected onto a horizontal plane, or onto a two-dimensional map of the territory corresponding to the position coordinates of the image points and the points uniquely associated with a corresponding component of the electricity distribution network and / or with an object in the environment surrounding the network are used to generate a representation of the at least one segment of the electricity network by means of a graph, in which graph the sets of image points classified as representing a support are represented by the nodes of the graph and the sets of image points classified as conductors are represented by arcs that connect at least two nodes together.

[0031] As regards the step of calculating the anomalies, in particular in relation to the distances between one or more components of the electricity distribution network and / or one or more objects of the surrounding environment and / or other types of interferences between the components and the objects of the surrounding environment, it is possible to use different algorithms for extracting geometric features from the map generated in the previous step.

[0032] The possible features that can be extracted are, by way of a non-limiting example: quality of the LIDAR data metrics; elevation profile; poles, for example their height and / or inclination, and features of the tower; cable extraction and corresponding measurements, such as the minimum distance of the cable from the road / ground, and others, for example; secondary substations; interferences with roads, waterways and buildings; interferences with vegetation, it being possible to calculate volumes of interfering vegetation and to identify trees that could hit the power lines in the event of a fall.

[0033] According to another aspect of the invention, the monitoring data of the electricity distribution network consist of a combination of photographic images thereof, optionally and preferably in high definition, and a 3D point cloud obtained by means of an acquisition technique of the so-called LIDAR type.

[0034] In this case, the data deriving from the photographic images and the data from the 3D point cloud can be processed according to one or more of the combinations or sub-combinations and / or variants described above.

[0035] Another embodiment of the present invention provides that, at the end of the processing of the monitoring data relating to the photographic images by means of the above-mentioned combination of computer vision algorithms according to the various models of specialization of the algorithms, the image data relating to the individual identified components of the electricity distribution network and / or to the individual objects of the surrounding environment are uniquely coupled or correlated with a corresponding support provided in the monitoring data relating to the 3D point cloud.

[0036] The aforementioned process is based on geometric data of the image data acquisition cameras and allows to prevent duplications relating to the identified components and / or identified anomalies, and also allows to define the exact position coordinates of the components.

[0037] In particular, according to an embodiment of the invention which is provided in combination with one or more of any of the previous characteristics and / or embodiments and / or variants, the method of the present invention provides a step of so-called three-dimensional (i.e. 3D) re-identification of the components present in the images, this step providing the correlation of the photographic images for which the position and the geometric orientation unit vectors of the cameras or of the image acquisition devices are known, with the 3D point cloud obtained by means of the LIDAR acquisition technique, or suchlike, and from which there is determined the geolocation of the corresponding components portrayed in the photographic images on the actual geographical position thereof within the LIDAR point cloud.

[0038] According to another characteristic, which can be provided in combination with one or more of any of the previous characteristics, combinations of said characteristics and / or sub-combinations thereof, the method can also provide, in combination with the photographic images of the electricity distribution network and / or with a representation thereof in the form of a cloud of points acquired by means of LIDAR or similar technologies, the use of satellite images.

[0039] As regards the satellite images, computer vision algorithms are provided which extract maps of the vegetation present along lines of the electricity distribution network from the images.

[0040] In one embodiment, the processing of the satellite images for mapping the vegetation can occur by means of a semantic segmentation algorithm.

[0041] The invention also concerns a system for implementing the method according to one or more of the embodiments described above.

[0042] In one embodiment, the system comprises: at least one monitoring data acquisition unit, according to at least one of the following acquisition technologies: acquisition of photographic images, acquisition of 3D point clouds by means of LIDAR techniques, acquisition of satellite images; at least one system for processing the monitoring data, comprising a generic processing hardware in which a program is loaded in which there are encoded the instructions to make the generic processing hardware capable of executing the steps of the method according to one or more of the embodiments and variants described above, and in particular the instructions for executing the algorithms provided for processing the monitoring data according to one or more of the embodiments and variants of the method described above.

[0043] According to one embodiment, the system comprises in combination one or more terminals for displaying and / or communicating the results of the processing of the monitoring data in the form of images and / or graphs and / or in the form of data expressed in alphanumeric format.

[0044] The advantages of the present invention are evident from what described above:

[0045] The method and system make it possible to plan virtual inspections efficiently, avoiding unnecessary travel on site to perform an assessment. Furthermore, the detection of anomalies and the management of vegetation by means of 3D clouds of LIDAR points and images lead to the ability to predict any situations that can cause damage to the network and / or require interruptions thereof in advance, having a positive impact on the quality of the service.

[0046] Using classified LIDAR data it is possible to detect anomalies related to unwanted vegetation, which contributes to the predictive maintenance use case described above. By combining cloud point data with satellite images, it is possible to massively collect vegetation mapping data (for example, tree cover density and dominant leaf type) which in turn makes even predictive vegetation mapping possible, and therefore a better management of maintenance activities relating to containing the vegetation mass and to maintaining the network’s passage corridors.

[0047] DESCRIPTION OF THE DRAWINGS

[0048] These and other characteristics and advantages of the present invention will become apparent from the following description of some embodiments, given as a non-restrictive example with reference to the attached drawings wherein:

[0049] - figure 1 shows an example of the acquisition method which uses a combination of photographic acquisition and acquisition of a set of 3D points by means of LIDAR technology.

[0050] - figure 2 shows the first and second step of the method according to the present invention, relating to the identification in an image of regions of interest containing one or more components of the electricity distribution network and to the recognition of the individual components of the electricity distribution network by means of computer vision, respectively.

[0051] - figure 3 shows an example of the third step of identifying, in a specific region of interest, a cutout containing a component of the electricity distribution network and the different computer vision algorithms specialized in recognizing the type of object, any anomalies and / or materials of the component shown.

[0052] - figure 4 shows the result of the image processing step which provides a semantic segmentation of the image pixels.

[0053] - figure 5 graphically shows the functionalities of the automatic step of determining image quality according to a progressive and continuous value scale of said quality.

[0054] - figure 6 shows a flow diagram of an example of the execution of the method according to the present invention. - figure 7 schematically shows what happens in the step of re-identification and mapping of a component recognized in the images, in the map of the 3D point cloud.

[0055] - figure 8 shows, by way of example and schematically, a three-dimensional reconstruction from a 3D point cloud acquired by means of LIDAR technique and from which the position and conformation of objects in the environment surrounding the network and its components can be extracted, for the purposes of extracting parameters such as distance and others.

[0056] - figure 9 shows the results of the semantic segmentation of the points of the 3D cloud obtained by means of LIDAR acquisition technique, relating to the identification of the components present in the electricity distribution network.

[0057] - figure 10 shows the results of the semantic segmentation of the 3D point cloud relating to the recognition of objects present in the surrounding environment which said electricity distribution network passes through.

[0058] We must clarify that the phraseology and terminology used in the present description, as well as the figures in the attached drawings also in relation as to how described, have the sole function of better illustrating and explaining the present invention, their purpose being to provide a non-limiting example of the invention itself, since the scope of protection is defined by the claims.

[0059] To facilitate comprehension, the same reference numbers have been used, where possible, to identify identical common elements in the drawings. It is understood that elements and characteristics of one embodiment can be conveniently combined or incorporated into other embodiments without further clarifications.

[0060] DESCRIPTION OF SOME EMBODIMENTS

[0061] Some embodiments described using the attached drawings concern a method for the analysis of monitoring data of electricity distribution networks for the identification of the components present in said network and any anomalies thereof.

[0062] With reference to fig. 1, 100 shows the map of a geographical area which a line of an electricity distribution network passes through, as shown by the image 101.

[0063] At least one, preferably more, photographic images, preferably in high definition, are acquired of the line and in particular of the pylon present in said geographical zone, as shown with 101. Thanks to a LIDAR type acquisition technology, the point cloud representing the line is also acquired, as shown with 102.

[0064] In accordance with the present invention, according to a first processing step, an algorithm is used for processing the image 101 that identifies, in the image 100, all the components represented in such image, and in particular the regions of interest of the image 100 in which the components are shown. These regions of interest are in the form of so-called “bounding boxes”, as indicated with 200 in fig. 2.

[0065] Once the regions of interest in the form of these bounding boxes have been determined, the image cutouts, such as the one 300 containing the image of at least one component of the electricity distribution network, are subjected to a combination of different computer vision algorithms that are specifically trained to be specialized in performing different tasks. 310 indicates the step of classifying the shape of the insulator shown in the cutout 300. 320 indicates that the cutout is subjected to processing by means of a computer vision algorithm, configured and / or calibrated to identify, or classify, an anomaly of the insulator, as indicated with 320.

[0066] Finally, 330 schematically shows a third processing mode of the image cutout 300, according to which a specialized computer vision algorithm processes the data of the cutout 300 in order to classify the type of material.

[0067] The processing of the cutout 300 may also be performed with a semantic segmentation algorithm, which leads to the result shown in another cutout 400.

[0068] Following at least one of the processing steps according to figs, from 2 to 4, and before proceeding to the additional processing step, the method provides a step of verifying the quality of the images of the cutouts 300. In this case, as shown in fig. 5, the quality is expressed in the form of a combination of parameters that in combination generate a quality index. These parameters can include image sharpness, contrast, resolution and one or more other parameters that may be necessary for specific factors and related to the zone in which the network is present.

[0069] As is evident, there is a quality parameter that goes from poor to the left of the sheet to good and excellent to the right of the sheet. In synchrony with this index, image cutouts 500 are shown, each visually showing an interpretation of the image quality.

[0070] The first two left cutouts are out of focus and therefore have a low quality. These cutouts 501, 502 are therefore eliminated from the monitoring data set, while the data relating to the image from 503 to 506 is kept.

[0071] Fig. 6 shows a complete example of an image processing process according to the present invention, the images 600 are subjected to a step of determining regions of interest, each of which can contain at least one component of interest, as indicated by box 601. The parts of the image relating to the zones of interest of which 603 represents an example (also referred to as ROI), are subjected to automatic quality control, as indicated by box 602.

[0072] The components of the various regions of interest are listed in the right column, in which boxes 604 and 605 show the results of a processing by means of a semantic segmentation algorithm that leads to the identification of the vegetation and to the verification of the existence of damage to the concrete. This processing is performed on the data relating to the region of interest 603.

[0073] The same image can also be subjected to processing by means of a classification algorithm, which in turn can provide information on the type of support in relation to the material and shape, as indicated by boxes 606 and 607.

[0074] As shown, the process is extended to all the regions of interest detected, as shown by the steps of activating the detectors 608 and 609 that lead to the identification of different types of components referred to in the lists corresponding to the two detectors 606 and 607 and, at the end, by applying classification and / or semantic segmentation algorithms 610, 611, 612, 613, 614, 615 that are specific to a certain type of component, different anomalies and / or different types of components and / or materials are recognized.

[0075] Also in this case, downstream of the detectors 606 and 607, before the application of the semantic segmentation and / or classification algorithms, an image quality verification step is applied to the image cutouts provided by said detectors 606, 607, which is performed as described above.

[0076] At the end of the processing process dedicated to the photographic images, and as shown in fig. 7, a step of re-identification and mapping is carried out, that is, of projecting onto the 3D map the cloud of points of components recognized in the processing of the photographic images. Fig. 7 shows, with 700, the result of the re-identification of the components recognized in the images also in the LIDAR point cloud. The image labeled with 710 shows the cutout of the photographic image containing the component to be mapped and which has been recognized thanks to the classification processing.

[0077] Fig. 7 shows, with 720, the image of the map containing the image of the component.

[0078] In this case, the position of the component of the map is incorrect, because the points in the positioning map following the re-identification are offset by an error of 36.81 meters for the top and 37.48 meters for the bottom.

[0079] According to one embodiment, the aforementioned step of 3D re-identification of the components of the electricity distribution network or of the objects of the environment surrounding it provides the correlation of the photographic images for which the position and the geometric orientation unit vectors of the cameras or of the image acquisition devices are known with the point cloud obtained by means of the LIDAR acquisition technique, or suchlike, and from which the geolocation of the corresponding components portrayed in the photographic images on the actual geographical position thereof within the LIDAR point cloud is determined.

[0080] Fig. 8 graphically shows an image reconstructed using the LIDAR point cloud.

[0081] The processing steps provide a semantic segmentation of the points and subsequently a classification in order to identify and recognize the position of the vegetation masses and their size, and the position of the buildings present in the external environment in which the electricity distribution network passes.

[0082] Fig. 9 shows some categories of components that have been recognized and classified by the semantic segmentation algorithm.

[0083] Fig. 10 shows a similar situation, but with reference to objects present in the surrounding environment and potentially interfering with the lines of the electricity distribution network.

[0084] Thanks to a feature extraction algorithm, from the data represented by images from 8 to 10 it is possible to extract indications relating to the distance of the vegetation or buildings from the power lines and assess their current or future danger.

[0085] Thanks to the above, and to the combination of a plurality of algorithms specialized in very limited tasks, that is, in the recognition of only one type of component and / or one type of material etc., it is possible to reduce the computational burden and automate the analysis of the monitoring data of an electricity distribution plant, with further advantages in terms of costs and implementation times.

[0086] Moreover, the virtualization of analysis operations makes it easy to generate a standardization that allows to overcome limitations caused by regional specifications that are different from region to region or geographical zone to geographical zone, making possible a remote control of networks that can possibly even bypass on site human intervention.

[0087] It is clear that modifications and / or additions of steps or parts may be made to the method and to the system for implementing such method as described heretofore, without thereby departing from the field and scope of the present invention, as defined by the claims.

[0088] It is also clear that, although the present invention has been described with reference to some specific examples, a person of skill in the art will be able to achieve other equivalent forms of method and system for implementing such method, having the characteristics as set forth in the claims and hence all coming within the field of protection defined thereby.

[0089] In the following claims, the sole purpose of the references in brackets is to facilitate their reading and they must not be considered as restrictive factors with regard to the field of protection defined by the claims.

Claims

CLAIMS1. Method for the analysis of monitoring data of electricity distribution networks for the identification of components present in said electricity distribution network and any anomalies thereof, characterized in that said method comprises the following steps: providing for the acquisition of monitoring data by means of at least one, preferably two or more techniques for acquiring said monitoring data; said at least one monitoring data acquisition technique providing for the interaction of one or more components present in said electricity distribution network and / or of one or more objects provided in an environment surrounding said components with electromagnetic radiation, in particular said at least one acquisition technique being of the photographic type or being a remote sensing technique of the so-called LIDAR (Laser Imaging Detection and Ranging) type, or consisting of satellite images; processing the monitoring data acquired with at least one of said acquisition techniques for the identification of said components present in said electricity distribution network and / or of said one or more objects present in the environment surrounding said electricity distribution network and determining, from said monitoring data, anomalies of one or more of said components, wherein the monitoring data collected according to at least one of said acquisition techniques are processed by means of combinations of classification algorithms and segmentation of the monitoring data, wherein the monitoring data consist of both photographic images, preferably in high definition, and also a 3D point cloud collected by means of LIDAR technology, or suchlike, which monitoring data are processed by means of machine learning algorithms specifically trained, respectively, for the processing of photographic images and for the processing of 3D point clouds acquired using LIDAR technology, or suchlike.

2. Method according to claim 1, wherein at least for the monitoring data consisting of photographic images, the method provides to subject said photographic images, or at least certain regions thereof, to an automatic evaluation of a quality of the photographic image.

3. Method according to claim 2, wherein it is provided to determine a thresholdvalue corresponding to a minimum threshold value of acceptable quality and to determine the same value for the image or images of the electricity distribution network to be processed, eliminating from the monitoring data to be processed, those relating to images whose image quality value is lower than that of the minimum quality threshold.

4. Method according to one or more of the preceding claims, which method provides to carry out the following steps of processing the monitoring data consisting of said photographic images: a first object detection step in which, by means of a first computer vision algorithm model, regions of interest are identified in the photographic images in which one or more components of the electricity distribution network are respectively represented; subsequently, in a second processing step, the various regions of interest identified in said first step, optionally only said regions of interest, are subjected to processing by means of one or more models of additional computer vision algorithms specialized for the identification of each component of the electricity distribution network which can potentially be provided in the electricity distribution network; at the end of this step, a third processing step is performed in which the photographic image is cropped, generating an image cutout for one or more of said components identified in the previous step and said image cutouts are subjected to one or more additional models of computer vision algorithms specialized, respectively, for the identification of anomalies of said components and / or of the materials which said components are made of and / or of optional further characteristics of said components identified in said photographic images.

5. Method according to claim 4, wherein said one or more additional models of computer vision algorithms, in particular those applied for processing said image cutouts, consist of classification and / or semantic segmentation algorithms, which are specialized, respectively, to identify the regions of interest in an image, i.e. the regions of the image that contain target objects, one or more components of the electricity distribution network, to recognize a typology of objects in each of said zones of interest, and to identify anomalies and / or materials and / or otherparameters relating to said components in cutouts of the image and of said regions of interest which contain said components.

6. Method according to claim 4 or 5, wherein before said first processing step and / or preferably before the second and / or preferably before the third processing step the regions of interest of the image and / or the cutouts of said regions of interest containing the image of one or more components of the electricity distribution network are subjected to a further processing to verify the image quality, wherein: quality parameters of the image are measured, in particular image definition and / or contrasts and / or sharpness and / or focus or optionally other parameters, and said values are compared with preset threshold values which define a minimum image quality threshold, the image cutouts for which the result of the comparison has revealed values of said parameters lower than said minimum image quality threshold being eliminated from the processing process of the step provided after said quality analysis.

7. Method according to one or more of the preceding claims, wherein the method provides that a 3D point cloud acquisition technique, such as said LIDAR type technology, or suchlike, is used as the monitoring data acquisition technique.

8. Method according to claim 7, wherein there is provided the succession of steps of processing the monitoring data consisting of said 3D point cloud, comprising: a first step of semantic segmentation of the 3D point cloud, each component being segmented with reference to its own geometry; a second step of identifying the components of the electricity distribution network and / or of objects present in the surrounding environment, such as vegetation, artefacts, such as cabins, buildings and other supports such as poles, or suchlike, and of the cables present in the electricity distribution network subjected to monitoring; the generation of a map of the electricity distribution network and / or of the surrounding environment; the calculation of anomalies relating to the distances between one or more components of the electricity distribution network and / or one or more objects of the surrounding environment and / or interference of other types between said components and said objects of the surrounding environment.

9. Method according to one or more of the preceding claims, wherein at the end of the processing of the monitoring data relating to the photographic images by means of said combination of computer vision algorithms according to said various models of specialization of said algorithms, the image data relating to said individual identified components of the electricity distribution network and / or to said individual objects of the surrounding environment are uniquely coupled or correlated with a corresponding support provided in the monitoring data relating to the 3D point cloud.

10. Method according to one or more of the preceding claims, wherein in combination with the photographic images of the electricity distribution network and / or with a representation thereof in the form of a 3D point cloud acquired by means of LIDAR or similar technologies, the use of satellite images is also provided, which are subjected to processing by means of computer vision and / or semantic segmentation algorithms which extract from said images maps of the vegetation present along lines of said electricity distribution network.

11. Method according to one or more of the preceding claims, which method comprises another step of so-called three-dimensional re-identification of the components or objects of the surrounding environment identified in the images, said step providing: the correlation of the photographic images for which the position and the geometric orientation unit vectors of the cameras or of the image acquisition devices are known with the 3D point cloud obtained by means of the LIDAR acquisition technique, or suchlike, and the geolocation of the corresponding components portrayed in the photographic images on an actual geographical position thereof within said 3D point cloud acquired by means of LIDAR technology.

12. System for implementing the method according to one or more of the preceding claims, characterized in that said system comprises: at least one monitoring data acquisition unit, according to at least one of the following acquisition technologies: acquisition of photographic images, acquisition of 3D point clouds by means of LIDAR techniques, acquisition of satellite images; at least one system configured for processing said monitoring data comprisinggeneric processing hardware in which a program is loaded in which there are encoded the instructions to make said generic processing hardware capable of executing the steps of the method according to one or more of the preceding claims, and in particular the instructions for executing the algorithms provided for processing said monitoring data according to a method in accordance with one or more of the preceding claims.

13. System according to claim 12, wherein one or more terminals are provided for displaying and / or communicating the results of said processing of the monitoring data in the form of images and / or graphs and / or in the form of data expressed in alphanumeric format.

Images