CN110135599B - Unmanned aerial vehicle electric power inspection point cloud intelligent processing and analyzing service platform - Google Patents
Unmanned aerial vehicle electric power inspection point cloud intelligent processing and analyzing service platform Download PDFInfo
- Publication number
- CN110135599B CN110135599B CN201910407334.3A CN201910407334A CN110135599B CN 110135599 B CN110135599 B CN 110135599B CN 201910407334 A CN201910407334 A CN 201910407334A CN 110135599 B CN110135599 B CN 110135599B
- Authority
- CN
- China
- Prior art keywords
- point
- point cloud
- segmentation
- data
- unmanned aerial
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C3/00—Registering or indicating the condition or the working of machines or other apparatus, other than vehicles
Abstract
The invention provides an unmanned aerial vehicle power inspection point cloud intelligent processing and analyzing service platform, which mainly comprises (a) an infrastructure layer; (b) a middleware layer; (c) an application layer; (d) a user layer; the specific construction process comprises (1) a platform frame overall architecture; (2) line patrol data storage and organization; (3) a line patrol data processing core algorithm; (4) line patrol data rendering technology and lean application. The advantages are that: the unmanned aerial vehicle laser radar, cloud service, real-time rendering, machine learning and other advanced technologies are comprehensively applied, related theories, algorithms and architecture systems are used as supports, an integrated solution under a cloud computing architecture is provided for the problems of the unmanned aerial vehicle power inspection point cloud in the aspects of data management, intelligent processing, lean application and the like, and open storage, automatic processing, intelligent analysis and lean application of unmanned aerial vehicle power inspection results are achieved.
Description
Technical Field
The invention relates to an unmanned aerial vehicle power inspection point cloud intelligent processing and analyzing service platform, and belongs to the technical field of power supply equipment maintenance and safety.
Background
The power transmission line is regularly inspected in a patrol mode, the running condition of the power transmission line, the change conditions of the surrounding environment of the power transmission line and the line protection area are mastered and known at any time, and the power transmission line inspection method is heavy daily work of power supply enterprises. The manual inspection is a traditional inspection mode, and due to the fact that the terrain environment of a power transmission line corridor is complex, in some regions with severe conditions such as crossing rivers or mountains and mountains, inspection roads are lacked in sections along the line, the labor intensity of the inspection mode is high, the working conditions are hard, and the running condition of the power transmission line cannot be mastered in time. Unmanned aerial vehicle patrols and examines as a kind and uses equipment of patrolling and examining such as laser radar, visible light and thermal infrared imager to carry out the whole new inspection technology of patrolling and examining to transmission line, has advantages such as swift rapidly, work efficiency is high, do not receive the region influence, patrol and examine high quality, security height, unmanned aerial vehicle's application is the effective solution of intelligent development of circuit patrol and examine, is particularly useful for the electric power industry, and electric power patrols and examines and has become one of the civilian fields of unmanned aerial vehicle that china has the development potentiality most.
With the annual popularization of unmanned aerial vehicles in power inspection application and the continuous compression of inspection periods, the annual increase of the inspection mileage of the unmanned aerial vehicles will exceed 10% in the next years. The task amount of processing and analyzing the point cloud data of the unmanned aerial vehicle power inspection is quite huge, and the advanced application of inspection result data is urgently needed by a lean management target. Therefore, the processing of the line patrol point cloud data has very wide market space and potential, and the intelligent processing and analysis service platform for the unmanned aerial vehicle power patrol point cloud is urgently needed to be researched and developed at present.
Utilize the lidar scanning system that unmanned aerial vehicle platform carried on, can acquire high-tension transmission line's high density, high accuracy lidar point cloud and optical image data fast, realize that the true three-dimensional scene of transmission line body and surrounding environment rebuilds fast, carry out high accuracy cubical space and survey, dangerous point detection, simulation operating mode analysis and passageway visual management etc.. Since the power inspection application popularization of the unmanned aerial vehicle is formally started in 2013, the technology of scanning the power transmission line by the airborne laser radar is deeply and widely applied. In the face of rapidly increasing market demands, a plurality of scientific research institutions and scientific companies actively participate in market layout, and a complete industrial chain covering branch fields of unmanned aerial vehicle design and manufacture, laser sensor development, flight inspection service, data processing and analysis service, software system development and the like is gradually formed. Industry application in recent years shows that compared with the traditional operation mode of manually inspecting the power transmission line, the unmanned aerial vehicle laser radar scanning has the obvious advantages of high inspection speed, high imaging precision, rich types of ground objects contained in data, flexibility, good safety and the like, and can quickly reconstruct the three-dimensional scene of the power transmission corridor. At present, a long-acting working mechanism for power transmission line inspection, which is mainly based on machine inspection and assisted by human inspection, is initially established by units such as national power grid companies and southern power grid companies.
However, even if the operation is performed by the machine patrol method, there are certain problems in the fields of data storage management, automated processing analysis, and intensive production data application. The laser point cloud data is high in density and large in size, the number of points between every two power transmission gears (the range between two adjacent power towers) is as large as ten million on average, and the high-resolution image data synchronously acquired by an onboard camera is added, so that the inspection original data volume is quite large, high requirements on computer hardware in the aspects of data storage, processing, analysis and the like are undoubtedly provided, and a power transmission and transportation inspection department is synchronously required to be equipped with a corresponding system operation and maintenance guarantee system. On the other hand, power transmission service analysis and application are developed based on point cloud data, and various complex data processing (noise filtering, navigation band data splicing and registration, spectrum and point cloud data fusion, point cloud segmentation, feature extraction, ground feature identification and the like) must be performed firstly, although many related algorithms are available at home and abroad, the algorithms are limited to specific data, specific scenes and specific ground feature types, the input required by the algorithms is ideal, and the scale of the processed data is limited, so that the robustness, universality and expandability of the algorithms are all to be improved, and the parameter setting and processing flows are different according to different requirements of users, so that the mainstream commercial processing tools and open source software packages (such as Terrasolide, INPHO, ENVI LiDAR, CloudCompare, PCL and the like) in the market at present are mainly assisted to manually classify or semi-automatically process human-computer interaction, and the operation efficiency of point cloud data processing cannot be effectively improved for a long time, greatly limiting the need for users to quickly acquire and use data achievements. In addition, in the aspect of analyzing routing inspection result data, due to different technical levels, implementation manners and operation habits of various manufacturers, the problems of low efficiency, low result quality, even incapability of meeting the technical regulations of the industry in the analysis process and the like often exist.
Disclosure of Invention
The invention aims to overcome the defects of an unmanned aerial vehicle power inspection operation mode and a data processing method in the existing power grid system, provides an unmanned aerial vehicle power inspection point cloud intelligent processing and analyzing service platform, comprehensively utilizes advanced technologies such as unmanned aerial vehicle laser radar, cloud service, real-time rendering and machine learning, takes relevant theories, algorithms and a framework system as supports, provides an integrated solution under a cloud computing framework aiming at the problems of the unmanned aerial vehicle power inspection point cloud in the aspects of data management, intelligent processing, lean application and the like, and realizes open storage, automatic processing, intelligent analysis and lean application of unmanned aerial vehicle power inspection results.
The technical solution of the invention is as follows: unmanned aerial vehicle electric power patrols and examines intelligent processing of point cloud and analysis service platform, its structure mainly includes following component:
(a) infrastructure layer: resources such as storage, calculation, safety, operation and maintenance management and the like required by the middle layer or the user are provided, and the resources are pooled through technologies such as virtualization and the like, so that the resources are allocated and rapidly deployed as required; including processing CPUs, memory, storage, networks and other basic computing resources and network security and operation and maintenance regulatory architectures.
(b) A middleware layer: the middleware cluster comprises a data storage module, a calculation service module, a load balancing module, a resource scheduling module, a security management module, a user management module, a charging management module and an operation and maintenance management module; the middle layer actually refers to a platform developed by software as a service, and the platform is submitted to a user in a mode of an application layer. In this framework, the middleware layer is actually present in the form of a software entity deployed into the infrastructure cluster. The system stores, dispatches and deploys a ROSEHA system of American ROSE data system company; the IBM xCAT cluster management system realizes security management, resource scheduling, user charging and operation and maintenance management.
(c) An application layer: providing various required application software and services for a user by a friendly user interface, directly facing to the requirements of the client, and providing inspection result storage service, point cloud data processing and analyzing service and inspection result application service for the client; the achievement storage service of the framework comprises: laser point cloud, channel images, asset accounts and defect hidden danger data. The point cloud data analysis service comprises: quality inspection, noise filtering, point cloud segmentation and classification, spectrum fusion, inter-file cutting, working condition analysis and line patrol report compiling. The inspection result application services comprise real-time rendering, query statistics, defect elimination management, history comparison, working condition simulation and point cloud micro-service.
(d) And (3) a user layer: also called service terminal, is an interactive interface facing the end user, providing service access and applications.
The construction process specifically comprises the following steps:
(1) the overall architecture of the platform framework is as follows: the method comprises the following steps of coupling line patrol point cloud and cloud computing, dynamically configuring resources according to user requirements by using a cloud computing technology, and providing synchronous or asynchronous computing services by using a large number of computing hosts to realize parallel synchronous real-time processing and analysis of super-large scale unmanned aerial vehicle patrol point cloud data;
(2) line patrol data storage and organization: the method comprises the steps that a power transmission line basic ledger is used as an index, a routing inspection result metadata model is used as a drive, virtualization of storage equipment is used as a basis, and powerful technical support is provided for uploading, accessing, managing and downloading of mass unmanned aerial vehicle routing inspection results through open cloud storage service; the routing inspection results in each power transmission and inspection department are organized and managed according to the hierarchical structure of line interval-data version-data category-data entity, so that the time for carrying out business work of power transmission and inspection team personnel is ensured, the management and use efficiency of routing inspection result data is improved, the waste of storage facility resources is effectively avoided, and the cost is saved for users;
(3) the line patrol data processing core algorithm:
firstly, point cloud self-adaptive segmentation: on the basis of in-depth research on heuristic segmentation, probabilistic model segmentation, graph optimization segmentation, parameter model optimization segmentation, machine learning/deep learning segmentation and other algorithms, under a Riemann geometric framework, a heuristic voxel segmentation and graph optimization method is integrated, and a layered optimization self-adaptive segmentation algorithm is designed;
the specific implementation is divided by coarse granularity andthe fine granularity segmentation comprises two steps: the coarse-grained segmentation input parameters are two: for a point p, in the point cloud, the point p is taken as a center, and a spherical neighborhood with the radius of Eps is the Eps neighborhood of the point p) and MinPts (a set threshold value is used for comparing the number of points in the Eps neighborhood of the point p), and clustering is realized according to the two parameters, and the points in the point cloud are divided into three categories: core point: points in the neighborhood of Eps that have a number of neighbors greater than MinPts; boundary points are as follows: points within the neighborhood of Eps that are less than MinPts and within the neighborhood of some core point Eps; noise points: points in the neighborhood of Eps that are less than MinPts and not in the neighborhood of some core point Eps. Then, the core points with the density being reachable are gathered into a class, and the density being reachable can be expressed as: given a point cloud data set, p1,p2......pnWherein p ═ p1，q＝pnIf point p isiTo pi+1(i ═ 1,2,3,. n), at point p, satisfying point pi+1In the neighborhood of Eps, at the same time point pi+1The number of neighbor points in the Eps neighborhood is greater than MinPts, then the point p to point q density is said to be reachable. And for the boundary point, the class of the boundary point is kept as same as that of a certain core point in the Eps neighborhood, the noise point is not classified, and the coarse-grained segmentation can be completed through the process.
After the coarse-grained segmentation is finished, fine-grained segmentation is further performed on each segmentation unit, so that the homogeneity of the segmentation units is ensured, and the method specifically comprises the following steps: segmenting unit data p for a set of coarse-grained three-dimensional point cloudsi∈R31, 2.. n, which are set to be classified into K classes (K < n), fine granularity segmentation attempts to find a labeling mode to minimize the sum of the distances from each point in the point cloud to the centroid point of the corresponding class, i:
wherein S iscSet of points belonging to the c-th category, pcIs ScThe center of mass of the lens.
Each point cloud cluster for coarse-grained segmentation
(d ═ 1,2, …, L) the following steps are performed:
(a) defining a currently input point cloud cluster as D, and selecting K points from the D as an initial centroid;
(b) calculating the classification of each point according to the distance, (t) represents the t-th iteration:
where min (-) represents the minimum in the set.
(c) Updating the centroid point for each category:
(d) repeating b) to c) until the centroid of all classes no longer changes, i.e. the class of each point no longer changes, i.e.:
(e) through the steps, the point cloud cluster
Clustering into K sub-point cloud clusters Sdc(c is 1,2, …, K), judging the number of points N in each sub-point cloud clusterdcIf it is greater than the threshold T. If N is presentdcT is less than or equal to T, then SdcAs output of the algorithmic over-segmentation. Otherwise, let D be SdcAs input, continuing to execute (a) - (d) to obtain a sub-point cloud cluster SdcK sub-point cloud clusters. And (4) carrying out the process aiming at each segmentation unit, and completing the segmentation of all point clouds.
② automatic multi-target classification of laser point cloud based on multi-scale dimension features, wherein the method comprises analyzing the core of various ground objects by using multi-scale point cloud classification algorithm based on three-dimensional geometric dimension featuresAnd (3) establishing a multi-scale classifier by using the dimensional characteristics of the point cloud data under different scales, automatically searching the optimal classification scale combination, and finally realizing multi-target automatic classification. The layering algorithm is implemented as follows: first, the point cloud is projected onto a two-dimensional plane, and then the two-dimensional plane is mapped to a two-dimensional manifold space
And finally, through controlling different thresholds of different levels and by means of a classical normaize cuts algorithm, obtaining segmentation results of objects of different levels and inheritance relations among the objects, and obtaining the multi-level point set object with content perception.
(4) Line patrol data rendering technology and lean application: optimizing a spatial index algorithm of the ultra-large-scale point cloud data, and realizing real-time rendering and interactive measurement of massive three-dimensional point cloud scenes by combining a WebGL technology on the basis of a modifiable nested octree data structure, so that long-time file copying among different hosts before a user browses data is avoided; services such as inquiry and statistics functions, defect elimination management functions, historical state comparison functions and the like of routing inspection results and line defect information are provided; the system provides simulation analysis services such as tree growth simulation, windage yaw working condition simulation, wire icing detection and the like, and simultaneously develops a point cloud micro-service interface of RESTful based on a standard open protocol.
The invention has the advantages that: (1) based on cloud storage service, a solution for effectively organizing and storing massive unmanned aerial vehicle inspection result data is provided, the current situations that data management is not standard and results are inconvenient to use are favorably changed, a transmission line operation and maintenance unit can conveniently and effectively master the line operation state, and the national transmission and power supply safety is ensured.
(2) The cloud data processing method has the advantages that cloud processing and analysis of unmanned aerial vehicle power inspection point cloud data are achieved, overall production efficiency of unmanned aerial vehicle power inspection is greatly improved, service pressure of a power transmission operation inspection unit is relieved, and the important function of finding line operation hidden dangers in time is achieved.
(3) The sharing access and the deep application of the super-large scale unmanned aerial vehicle inspection point cloud result in the power transmission operation and inspection unit are realized, the application value of unmanned aerial vehicle power inspection result data is fully developed, and the lean management level of a power company is improved.
(4) Insisting on the drive of technical innovation and service innovation, a series of industrialized achievements with revolutionary significance for the whole unmanned aerial vehicle inspection industry are formed, and positive and profound demonstration effects are generated for the continuous healthy development of the unmanned aerial vehicle inspection industry.
Drawings
FIG. 1 is a schematic structural diagram of an unmanned aerial vehicle power inspection point cloud intelligent processing and analyzing service platform.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings.
The general architecture of the unmanned aerial vehicle power inspection point cloud intelligent processing and analyzing service platform is shown in figure 1. The platform is divided into four organically connected logical levels from bottom to top by taking a policy and regulation and information security system and a customer service and operation and maintenance management system as guarantee:
(a) infrastructure layer: the infrastructure as a service (IaaS) is to provide resources such as storage, computation, security, operation and maintenance management required by a middle layer or a user, and to pool the resources by using technologies such as virtualization, so as to realize the on-demand allocation and rapid deployment of the resources.
(b) A middleware layer: platform as a service (PaaS), which is a layer developed from the top, provides services for users on the basis of resources provided by an infrastructure layer, and includes middleware clusters such as data storage, computing services, load balancing, resource scheduling, security management, user management, billing management, operation and maintenance management, and the like.
(c) An application layer: the software as a service (SaaS) is characterized in that a friendly user interface is used for providing various required application software and services for a user, an application layer directly faces to the requirements of the client, and enterprise-level applications such as inspection result storage service, point cloud data processing and analyzing service, inspection result application service and the like are provided for the client.
(d) And (3) a user layer: also called service terminal, is an interactive interface facing the end user, providing service access and applications.
(1) The overall architecture of the platform framework is as follows:
at present, in the aspect of processing and analyzing point cloud data of electric power inspection of an unmanned aerial vehicle, the domestic market mainly depends on running a stand-alone professional software tool on a workstation and adopts a man-machine interaction mode to operate. Due to the huge amount of point cloud data, the high price of a single workstation and the limited computing resources, the low working efficiency is inevitably caused. The cloud computing is a new computing mode capable of dynamically configuring resources according to user requirements, has super-strong elastic computing capability, and can provide synchronous or asynchronous computing services by simultaneously utilizing a large number of computing hosts to realize parallel synchronous real-time processing and analysis of point cloud data of the ultra-large scale unmanned aerial vehicle. According to the invention, line patrol point clouds and cloud computing are innovatively coupled, through an efficiency comparison test, a mainstream software tool in the market at present adopts manual interaction classification processing and line patrol analysis report compiling, the average time of each kilometer of a power transmission line is at least 6 hours, the full-automatic cloud parallel computing is adopted for 10 minutes at most, and the production efficiency can be improved by more than 30 times.
(2) Line patrol data storage and organization mode:
at present, in the aspect of storage of unmanned aerial vehicle inspection results, transmission and inspection departments in various regions are still at a lower management level, and a good management mechanism is not established. A large number of inspection results are stored sporadically on the workstations. In the aspect of the organization of the unmanned aerial vehicle inspection result, a directory or a file is mainly used as an organization structure, the name of the unmanned aerial vehicle inspection result is used as an index mark, a unified naming specification is lacked, and a great deal of inconvenience is brought to the classification, the retrieval and the use of the inspection result. Because the original routing inspection data is big data, the capacity of the existing storage equipment needs to be expanded every two years, and because professional and reasonable IT resource planning capacity is lacked, the hardware resource is often seriously wasted. By means of good opportunity of national broadband network speed increasing and cost reducing, the platform creatively takes a transmission line basic ledger as an index, a routing inspection result metadata model as a drive, a storage device virtualization as a basis and through open cloud storage service, powerful technical support is provided for uploading, accessing, managing and downloading of massive unmanned aerial vehicle routing inspection results. Under the service mode, the routing inspection results in each power transmission and transportation inspection department are organized and managed according to a layered structure of 'line interval-data version-data category-data entity', a user does not need to consider the problems of purchasing and operation and maintenance of storage equipment and the place where the data is stored, and can enjoy convenient data storage management service by paying corresponding service cost according to actual memory space, so that the time of carrying out business work by power transmission and transportation inspection team personnel is ensured, the management and use efficiency of routing inspection result data is improved, the waste of storage facility resources is effectively avoided, and the cost is saved for the user.
(3) The line patrol data processing core algorithm:
for a long time, factors such as research lag of related algorithms, unreasonable operation flow, high requirement on professional skills of operators and the like directly restrict the improvement of the point cloud data processing and analyzing efficiency. Currently, mainstream point cloud processing software in the market has certain automatic processing capability and classification accuracy on single ground objects such as noise filtering, ground point classification, wires, towers and the like, but has weak processing capability on other ground objects such as buildings, traffic facilities, vegetation and the like, and cannot execute multi-target classification processing tasks in parallel. In addition, the existing analysis mode mainly analyzes the tree line distance, the refinement degree is not high, and the practical value is low. The invention realizes the following point cloud processing and analysis in the power patrol of the unmanned aerial vehicle: firstly, point cloud self-adaptive segmentation. The point cloud segmentation is to divide the point cloud according to the geometric and texture characteristics of ground objects, so that the point clouds belonging to the same segmentation unit have similar characteristics, and the effective segmentation of the point cloud data is the premise of realizing high-level characteristics and semantic description. On the basis of in-depth research on algorithms such as heuristic segmentation, probability model segmentation, graph optimization segmentation, parameter model optimization segmentation, machine learning/deep learning segmentation and the like, a hierarchical optimization adaptive segmentation algorithm is designed by integrating heuristic voxel segmentation and graph optimization methods under a Riemann geometric framework, and the algorithm has clear logic, is easy to realize, has high operation efficiency and high fidelity of local details of ground objects, and lays an important foundation for multi-target full-automatic classification of three-dimensional point clouds. And secondly, automatically classifying the laser point cloud by multiple targets based on the multi-scale dimensional characteristics. We developed a multi-scale point cloud classification algorithm based on three-dimensional geometric dimension features. Firstly, the dimensional characteristics of various ground object core point cloud data under different scales are analyzed, then a multi-scale classifier is constructed, the optimal classification scale combination is automatically searched, and finally multi-target automatic classification is realized. Through verification, the method i) is efficient and can be applied to processing large-scene point cloud data; ii) high precision, the classification precision is better than 97.4%; and iii) the robustness is strong, the method can be used for processing complex scanning scenes, the complexity is expressed in data, point cloud missing, shielding and density non-homogeneous point clouds can be processed, the point cloud missing, shielding and density non-homogeneous point clouds are expressed in ground object scene categories, and multiple types of ground objects in the complex scenes can be segmented in a self-adaptive mode. And the intelligent real-time working condition safety distance analysis is realized. By combing the related technical regulations of power transmission line design and operation, various dangerous influence factors are arranged into a quantitative rule base, the threshold range of each detection rule is defined, by means of an efficient spatial index algorithm and a clustering algorithm, not only can the dangerous and hidden dangers outside the line be accurately analyzed and checked, but also the condition that the power transmission line crosses and spans high railways and highways can be automatically checked, the method can intelligently and accurately measure the parameter indexes of the drainage wire to the tower minimum clearance distance, the drainage wire length, the wire maximum sag, the wire length, the wire grade distance, the wire split line distance and the like which need to be measured on site when the transmission line is handed over and accepted, and the analysis results are automatically output as report texts in a standard format in a chart form, so that the method has the ultra-strong analysis capability and greatly saves the workload of manually compiling analysis reports. Because the point cloud data is uniformly stored in the cloud end, the whole processing and analyzing process is completed in the cloud end computing, and the efficiency of the point cloud data processing and analyzing is improved in a breakthrough manner due to the strong cloud computing performance.
(4) Line patrol data rendering technology and lean application:
at present, the storage and management mode of the point cloud result of the unmanned aerial vehicle power inspection is lagged behind, the shared use and the service application of the point cloud data result in a power transmission and operation inspection department are limited to a great extent, the application value of the data result is not fully mined, and the overall utilization efficiency of data resources is not high. Aiming at the problem, on the basis of cloud storage management and cloud processing analysis, a spatial index algorithm of super-large-scale point cloud data is optimized, real-time rendering and interactive measurement of massive three-dimensional point cloud scenes are realized on the basis of a modifiable nested octree data structure and in combination with a WebGL (Web Graphics Library, WebGL) technology, and long-time file copying among different hosts before a user browses data is avoided; services such as inquiry and statistics functions, defect elimination management functions, historical state comparison functions and the like of routing inspection results and line defect information are provided; the simulation analysis services of tree growth simulation, windage yaw working condition simulation, wire icing detection and the like are provided. In addition, in order to facilitate integration with the existing business application system of a user, a RESTful point cloud micro-service interface is developed based on a standard open protocol. The unmanned aerial vehicle power inspection result application service framework belongs to the first time in China, and the demand of lean management of a power transmission line can be effectively met.
The invention provides an open cloud data storage service, uploads increasingly-growing and massive unmanned aerial vehicle power inspection results to the cloud as required for unified storage, can conveniently manage, retrieve and download historical inspection result data, and provides a cloud data processing and analyzing service with high throughput and high efficiency; the method redefines human-computer interaction from the perspective of specialized data processing, realizes a highly-automated and intelligent service mode of a point cloud data factory, enables operators to thoroughly get rid of complicated processing steps, truly realizes a one-click processing flow, and constructs an unmanned aerial vehicle power inspection point cloud result application service framework.
Claims (1)
1. Unmanned aerial vehicle electric power patrols and examines intelligent processing of point cloud and analysis service platform, its characterized by structure includes following component:
(a) infrastructure layer: providing storage, calculation, safety and operation and maintenance management resources required by the middle layer or a user, and pooling the resources through a virtualization technology to realize the allocation and rapid deployment of the resources as required, wherein the resource allocation and rapid deployment comprise a processing CPU, a memory device, a storage device, a network device, a basic calculation resource and a network safety operation and maintenance supervision system;
(b) an intermediate layer: the middleware cluster comprises a data storage module, a calculation service module, a load balancing module, a resource scheduling module, a security management module, a user management module, a charging management module and an operation and maintenance management module;
(c) an application layer: providing various required application software and services for a user by a friendly user interface, directly facing to the requirements of the client, and providing inspection result storage service, point cloud data processing and analyzing service and inspection result application service for the client;
(d) and (3) a user layer: the service terminal is an interactive interface which faces to the end user and provides service access and application;
the construction process specifically comprises the following steps:
(1) the overall architecture of the platform framework is as follows: coupling the inspection point cloud and cloud computing, dynamically configuring resources according to user requirements by using a cloud computing technology, and providing synchronous or asynchronous computing services by using a large number of computing hosts to realize parallel synchronous real-time processing and analysis of inspection point cloud data of the ultra-large scale unmanned aerial vehicle;
(2) and (3) routing inspection data storage and organization: the method comprises the steps that a basic power transmission line ledger is used as an index, a routing inspection result metadata model is used as a drive, virtualization of storage equipment is used as a basis, and technical support is provided for uploading, accessing, managing and downloading of routing inspection results of the unmanned aerial vehicle through open cloud storage service; the routing inspection results in each transmission operation inspection department are organized and managed according to a hierarchical structure of line interval-data version-data category-data entity;
(3) routing inspection data processing core algorithm:
firstly, point cloud self-adaptive segmentation: on the basis of in-depth research of heuristic segmentation, probability model segmentation, graph optimization segmentation, parameter model optimization segmentation and machine learning/deep learning segmentation algorithms, under a Riemann geometric framework, a heuristic voxel segmentation and graph optimization method is integrated, and a layered optimization self-adaptive segmentation algorithm is designed;
the specific implementation comprises two steps of coarse grain size segmentation and fine grain size segmentation: the coarse-grained segmentation input parameters are two: eps, namely the Eps neighborhood: for the point p, in the point cloud, taking the point p as a center, and a spherical neighborhood with Eps as a radius is an Eps neighborhood of the point p; and MinPts, a set threshold, for comparing with the number of points in the Eps neighborhood of the point p, and implementing clustering according to these two parameters, which will classify the points in the point cloud into three categories:
core point: points in the neighborhood of Eps that have a number of neighbors greater than MinPts;
boundary points are as follows: points within the neighborhood of Eps that are less than MinPts and within the neighborhood of some core point Eps;
noise points: points within the neighborhood of Eps that are less than MinPts and are not within the neighborhood of some core point Eps;
then, the core points with the accessible density are gathered into a class, and the accessible density is expressed as: given a point cloud data set, p1,p2......pnWherein p ═ p1，q＝pnIf point p isiTo pi+1N, i ═ 1,2,3,. n, and satisfies point p at point pi+1In the neighborhood of Eps, at the same time point pi+1If the number of neighbor points in the Eps neighborhood is greater than MinPts, the density from point p to point q is called to be reachable; for the boundary point, the class of the boundary point is kept as same as that of a certain core point in an Eps neighborhood, the noise points are not classified, and the coarse-grained segmentation can be completed in the process;
after the coarse-grained segmentation is finished, fine-grained segmentation is further performed on each segmentation unit, so that the homogeneity of the segmentation units is ensured, and the method specifically comprises the following steps: segmenting unit data p for a set of coarse-grained three-dimensional point cloudsi∈R31,2, n, which are set to be classified into K classes, K<n, then the fine granularity partition tries to find a labeling mode to minimize the sum of squared distances from each point in the point cloud to the centroid point of the corresponding category, namely:
wherein S iscSet of points belonging to the c-th category, pcIs ScThe center of mass of;
each point cloud cluster for coarse-grained segmentation
d is 1,2, …, L, the following steps are performed:
(a) defining a currently input point cloud cluster as D, and selecting K points from the D as an initial centroid;
(b) calculating the classification of each point according to the distance, (t) represents the t-th iteration:
wherein min (-) represents the minimum value in the set;
(c) updating the centroid point for each category:
(d) repeating (b) to (c) until the centroids of all classes no longer change, i.e. the class of each point no longer changes, i.e.:
(e) through the steps, the point cloud cluster
Clustering into K sub-point cloud clusters SdcC is 1,2, …, K, and the number of points N in each sub-point cloud cluster is judgeddcWhether it is greater than a threshold T; if N is presentdcT is less than or equal to T, then SdcAs an output of the algorithm over-segmentation; otherwise, let D be SdcAs an input, a relayContinuing to execute the steps (a) - (d) to obtain a sub-point cloud cluster SdcK sub-point cloud clusters of (1); the above process is implemented for each segmentation unit, namely the segmentation of all point clouds is completed;
② automatic multi-target classification of laser point cloud based on multi-scale dimension features comprises analyzing dimension features of core point cloud data of ground, vegetation, building, tower and transmission line under different scales by using a multi-scale point cloud classification algorithm based on three-dimensional geometry dimension features, constructing a multi-scale classifier, automatically searching the optimal classification scale combination, and realizing multi-target automatic classification
Firstly, constructing a Voronoi diagram based on gravity center constraint on an M space to obtain a point set object with content perception, and finally, obtaining segmentation results of objects in different layers and inheritance relationships among the objects by controlling different thresholds in different layers and by means of a classical normaize cuts algorithm to obtain a multi-layer point set object with content perception;
(4) routing inspection data rendering technology and lean application: optimizing a spatial index algorithm of the ultra-large-scale point cloud data, and realizing real-time rendering and interactive measurement of massive three-dimensional point cloud scenes by combining a WebGL technology on the basis of a modifiable nested octree data structure, so that long-time file copying among different hosts before a user browses data is avoided; providing inquiry and statistics functions, defect elimination management functions and historical state comparison function services of routing inspection results and line defect information; the system provides tree growth simulation, windage yaw working condition simulation and wire icing detection simulation analysis services, and simultaneously develops a point cloud micro-service interface of RESTful based on a standard open protocol.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910407334.3A CN110135599B (en) | 2019-05-15 | 2019-05-15 | Unmanned aerial vehicle electric power inspection point cloud intelligent processing and analyzing service platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910407334.3A CN110135599B (en) | 2019-05-15 | 2019-05-15 | Unmanned aerial vehicle electric power inspection point cloud intelligent processing and analyzing service platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110135599A CN110135599A (en) | 2019-08-16 |
| CN110135599B true CN110135599B (en) | 2020-09-01 |
Family
ID=67574609
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910407334.3A Active CN110135599B (en) | 2019-05-15 | 2019-05-15 | Unmanned aerial vehicle electric power inspection point cloud intelligent processing and analyzing service platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110135599B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674763B (en) * | 2019-09-27 | 2022-02-11 | 国网四川省电力公司电力科学研究院 | Transmission channel tower image identification method and system based on symmetry inspection |
| CN111352750B (en) * | 2020-03-04 | 2023-08-18 | 云南电网有限责任公司电力科学研究院 | Method and system for identifying defect hidden trouble of multi-source image of power transmission line |
| CN111681318B (en) * | 2020-06-10 | 2021-06-15 | 上海城市地理信息系统发展有限公司 | Point cloud data modeling method and device and electronic equipment |
| CN112508264A (en) * | 2020-12-02 | 2021-03-16 | 国网冀北电力有限公司经济技术研究院 | Method for planning path of big data of power transmission line stock project by using genetic algorithm |
| CN112465075B (en) * | 2020-12-31 | 2021-05-25 | 杭银消费金融股份有限公司 | Metadata management method and system |
| CN113134230B (en) * | 2021-01-08 | 2024-03-22 | 成都完美时空网络技术有限公司 | Clustering method and device for virtual objects, storage medium and electronic device |
| CN112883845B (en) * | 2021-02-02 | 2022-06-07 | 贵州电网有限责任公司 | Automatic pole tower type identification method based on laser LiDAR point cloud |
| CN113222170B (en) * | 2021-03-30 | 2024-04-23 | 新睿信智能物联研究院(南京)有限公司 | Intelligent algorithm and model for AI collaborative service platform of Internet of things |
| CN113688282A (en) * | 2021-07-23 | 2021-11-23 | 北京三快在线科技有限公司 | Data processing method and device, electronic equipment and readable storage medium |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102436654B (en) * | 2011-09-02 | 2013-07-10 | 清华大学 | Adaptive segmentation method of building point cloud |
| US20140280964A1 (en) * | 2013-03-15 | 2014-09-18 | Gravitant, Inc. | Systems, methods and computer readable mediums for implementing cloud service brokerage platform functionalities |
| CN105187771A (en) * | 2015-07-31 | 2015-12-23 | 山东创德软件技术有限公司 | Plant-level comprehensive supervision platform |
| CN106790583B (en) * | 2016-12-28 | 2020-07-10 | 江苏盛世华安互联网科技股份有限公司 | City internet of things management system based on cloud platform |
| CN107085597A (en) * | 2017-03-30 | 2017-08-22 | 浙江精工钢结构集团有限公司 | A kind of storage of BIM models and browsing environment building method based on cloud framework |
| CN109188459B (en) * | 2018-08-29 | 2022-04-15 | 东南大学 | Ramp small obstacle identification method based on multi-line laser radar |
-
2019
- 2019-05-15 CN CN201910407334.3A patent/CN110135599B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110135599A (en) | 2019-08-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110135599B (en) | Unmanned aerial vehicle electric power inspection point cloud intelligent processing and analyzing service platform | |
| CN106709067B (en) | Multisource heterogeneous space data circulation method based on Oracle database | |
| US11720606B1 (en) | Automated geospatial data analysis | |
| CN106651188A (en) | Electric transmission and transformation device multi-source state assessment data processing method and application thereof | |
| CN109493119B (en) | POI data-based urban business center identification method and system | |
| CN112559534B (en) | Remote sensing image data filing management system and method | |
| CN110021072B (en) | Holographic mapping-oriented multi-platform point cloud intelligent processing method | |
| CN106202335A (en) | A kind of big Data Cleaning Method of traffic based on cloud computing framework | |
| CN115587399A (en) | Project progress management system and application based on BIM | |
| Song et al. | Classifying 3D objects in LiDAR point clouds with a back-propagation neural network | |
| CN113570275A (en) | Water resource real-time monitoring system based on BIM and digital elevation model | |
| Ortega et al. | Generating 3D city models from open LiDAR point clouds: Advancing towards smart city applications | |
| CN114357694A (en) | Transformer substation digital twinning method and device based on large-scale point cloud | |
| Pukkala | Optimized cellular automaton for stand delineation | |
| Singer et al. | Dales objects: A large scale benchmark dataset for instance segmentation in aerial lidar | |
| CN108898244A (en) | A kind of digital signage position recommended method coupling multi-source element | |
| CN116796083B (en) | Space data partitioning method and system | |
| CN117371949A (en) | Three-dimensional visual model-based power transmission line construction safety monitoring method and system | |
| Xiong et al. | Novel ITS based on space-air-ground collected big-data | |
| Zhu et al. | Rural road network planning based on 5g and traffic big data | |
| Jung et al. | Development of Information Technology Infrastructures through Construction of Big Data Platform for Road Driving Environment Analysis | |
| Inoue et al. | Visualization of 3D cable between utility poles obtained from laser scanning point clouds: a case study | |
| Xu et al. | An improved algorithm for clustering uncertain traffic data streams based on Hadoop platform | |
| He et al. | Density-Based Road Segmentation Algorithm for Point Cloud Collected by Roadside LiDAR | |
| Liu et al. | SS-IPLE: Semantic segmentation of electric power corridor scene and individual power lines extraction from UAV-based Lidar point cloud |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20190816 Assignee: Nanjing xingyutu Information Technology Co.,Ltd. Assignor: NANJING FORESTRY University Contract record no.: X2020320000227 Denomination of invention: UAV power inspection point cloud intelligent processing and analysis service platform Granted publication date: 20200901 License type: Common License Record date: 20201120 Application publication date: 20190816 Assignee: Nanjing Wenjing Information Technology Co.,Ltd. Assignor: NANJING FORESTRY University Contract record no.: X2020320000230 Denomination of invention: UAV power inspection point cloud intelligent processing and analysis service platform Granted publication date: 20200901 License type: Common License Record date: 20201120 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20190816 Assignee: NANJING CANYON INFORMATION TECHNOLOGY Co.,Ltd. Assignor: NANJING FORESTRY University Contract record no.: X2020320000232 Denomination of invention: UAV power inspection point cloud intelligent processing and analysis service platform Granted publication date: 20200901 License type: Common License Record date: 20201123 |