CN111192363A - Cloud computing-based user power distribution room design generation method - Google Patents
Cloud computing-based user power distribution room design generation method Download PDFInfo
- Publication number
- CN111192363A CN111192363A CN201911341997.6A CN201911341997A CN111192363A CN 111192363 A CN111192363 A CN 111192363A CN 201911341997 A CN201911341997 A CN 201911341997A CN 111192363 A CN111192363 A CN 111192363A
- Authority
- CN
- China
- Prior art keywords
- power distribution
- design
- distribution room
- cloud
- neural network
- 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.)
- Granted
Links
- 238000013461 design Methods 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 49
- 239000011159 matrix material Substances 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 16
- 238000005516 engineering process Methods 0.000 claims description 13
- 238000011960 computer-aided design Methods 0.000 claims description 10
- 238000013507 mapping Methods 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 239000013598 vector Substances 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 5
- 230000001360 synchronised effect Effects 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 238000004088 simulation Methods 0.000 claims description 4
- 230000002146 bilateral effect Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 abstract description 7
- 238000011161 development Methods 0.000 abstract description 4
- 238000004519 manufacturing process Methods 0.000 abstract description 4
- 238000013480 data collection Methods 0.000 description 5
- 238000012938 design process Methods 0.000 description 3
- 238000009434 installation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009897 systematic effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Remote Sensing (AREA)
- Strategic Management (AREA)
- Primary Health Care (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a user power distribution room design generation method based on cloud computing, and belongs to the technical field of business expansion engineering. The currently adopted business expansion design method consumes more time to complete the design work and has low efficiency. The method comprises the steps of collecting three-dimensional information and pictures of a power distribution room through laser depth sensor data, obtaining parameters required by drawing by using a convolutional neural network, and transmitting the parameters to a CAD module; finally, outputting a practically available power distribution room design drawing; the method can effectively shorten the design time of the business expansion and improve the design efficiency of the business expansion. The invention can shorten the design period, standardize the engineering standard, increase the supply and marketing, improve the user satisfaction, finally continuously promote the generation of enterprise value and form a business mode of virtuous circle development by organically integrating power supply service, electric power engineering design, electric power engineering construction, electric power equipment manufacturing and the like.
Description
Technical Field
The invention relates to a user power distribution room design generation method based on cloud computing, and belongs to the technical field of business expansion engineering.
Background
The high-voltage expansion project is generally divided into three expansion types of temporary expansion, new high-voltage installation and high-voltage expansion, wherein the high-voltage expansion needs to collect data of high-voltage distribution equipment installed at a user end on site, the current work flow is that after a customer manager has communicated the demand, a designer finishes on-site checking work, collects data of a power distribution room and equipment of a user, and a draft drawing method is mainly adopted for photographing and power distribution room measurement. After the field collection is finished, the drawn sketch is redrawn in CAD software, and then the equipment design to be installed and the design of the existing collection are combined according to the requirement of the user business expansion to form a new power distribution room design drawing. After the drawing is confirmed by a customer manager and a user, the designer draws the construction drawing again and gives the construction drawing to an implementer to complete equipment installation work.
The current business expansion design method can consume about 1 week to complete the design work under all smooth conditions. The process consumes too much time for the grid, design, construction and the user himself, so that new technology needs to be introduced to shorten the work process.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a method for collecting three-dimensional information and pictures of a power distribution room through laser depth sensor data, acquiring parameters required by drawing by using a convolutional neural network, and transmitting the parameters to a CAD module; finally, an actually available power distribution room design drawing is output, and the power distribution room design generation method based on cloud computing can effectively shorten the power expansion design time and improve the power expansion design efficiency.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a method for generating a user power distribution room design based on cloud computing,
which comprises the following steps:
firstly, establishing an interface program WSGI (Web service gateway initiative) which adopts a Web server and a Web application program defined for python programming voice and a depth calculation framework;
secondly, collecting three-dimensional information of the power distribution room by using a laser depth camera through staff on the site of the power distribution room;
thirdly, transmitting the three-dimensional reconstruction information and the laser depth photo of the equipment to a WSGI of a cloud server through a wireless network;
fourthly, transmitting data to a convolutional neural network inference module based on a GPU cluster through a WSGI (wireless sensor and antenna array) interface of the cloud end, acquiring parameters required by drawing, and transmitting the parameters to a python API (application programming interface) interface of a CAD (computer-aided design) module;
and fifthly, using a cloud accelerated computing architecture to accelerate the space computation of data collected by the field sensor, and finally outputting a standard drawing of field collection of the power distribution room.
The method comprises the steps of collecting three-dimensional information and pictures of a power distribution room through laser depth sensor data, obtaining parameters required by drawing by using a convolutional neural network, and transmitting the parameters to a CAD module; finally, outputting a practically available power distribution room design drawing; the method can effectively shorten the design time of the business expansion and improve the design efficiency of the business expansion.
By applying the scheme of the invention, the existing high-voltage business expansion design process can be reduced from the original 5 working days to 0.5 working day, and the revolutionary improvement of the working efficiency is realized. The invention can organically integrate power supply service, power engineering design, power engineering construction, power equipment manufacturing and the like, shorten the design period, standardize the engineering standard, increase the supply and marketing, improve the user satisfaction degree, finally continuously promote the generation of enterprise value and form a business mode of virtuous circle development.
As a preferable technical measure:
the input data of the convolutional neural network is a distribution room three-dimensional reconstruction point cloud collected by a laser depth camera through a synchronous positioning and mapping technology SLAM, the point cloud applies a partial differential algorithm to obtain a point with large variation as a characteristic point cloud, and then a plurality of coordinate points are randomly extracted to serve as an input source of the convolutional neural network; the device image taken with the laser depth camera is used as another input to the convolutional neural network.
The three-dimensional data collection is realized through the synchronous positioning and mapping technology, the three-dimensional data collection can be combined with the field three-dimensional data collection and can be independently applied, the data collection efficiency can be effectively improved, and meanwhile, the diversity and the accuracy of input data of the convolutional neural network are improved.
As a preferable technical measure:
the training of the convolutional neural network is completed by a simulation system, and a three-dimensional engine for robot training, a gazebo, is adopted to randomly generate a plurality of power distribution rooms and equipment with different sizes; a virtual laser depth camera of the robot is used as sensor input in the system, the virtual laser depth sensor is installed on the body of the virtual robot, a robot SLAM technology is operated in an ROS (convolutional systematic) system to be used for autonomous navigation of the robot to collect space and equipment data, a data set for convolutional neural network training is automatically generated, and a pre-training network is generated through a GPU cluster; finally, design survey data, including the number of devices, the type of devices, and the coordinates of the devices relative to the three-dimensional reconstruction, are output through the pre-trained network.
As a preferable technical measure:
the SLAM technology aims to construct a surrounding environment map in real time according to virtual laser depth sensor data without any prior knowledge, and simultaneously presumes the self positioning according to the map, and a virtual robot carries a virtual laser depth sensor to move in an unknown environment;
for convenience, the movement of a continuous time is changed into discrete time t 1, … k, and at these times, the position of the virtual robot is represented by x, and then the position of each time is represented as x1, x2 … xk, which constitutes the track of the virtual robot;
in the aspect of a map, the map consists of a plurality of three-dimensional point coordinates, and at each moment, the virtual laser depth sensor can measure a part of the three-dimensional point coordinates to obtain observation data of the three-dimensional point coordinates;
the three-dimensional point coordinates are N in number and are represented by y1 and y2 … yn; the positioning problem x and the mapping problem y are solved by the motion measurement u and the virtual laser depth sensor reading z.
As a preferable technical measure:
calling a laser ToF sensor, finishing three-dimensional point cloud collection of the power distribution room by staff on the site of the power distribution room, and judging the collected laser point cloud through a simple quantity;
the point cloud with the point number exceeding 50000 can be used as the input of a room key point coordinate extraction convolutional neural network, and no more than 50000 points require a worker to collect the three-dimensional point cloud again;
after the point cloud is input into the convolutional neural network, the coordinates of the key points of the basic shape of the room are inferred (reference) and vectorized.
Through the on-site collection of the three-dimensional data, the three-dimensional data can be combined with a synchronous positioning and mapping technology, and can be independently applied, so that the data collection efficiency can be effectively improved, and meanwhile, the diversity and the accuracy of input data of the convolutional neural network are improved.
As a preferable technical measure:
the method comprises the steps that field workers shoot depth images of existing high-voltage distribution equipment of a user by using the same laser ToF sensor, and the obtained laser depth images firstly pass through an invalid pixel point judgment process;
if the number of the invalid pixel points exceeds 15% of the total number of the pixel points, the worker is required to shoot the high-voltage distribution equipment again. And (3) inputting the depth image with invalid pixel points less than 15% of the total pixel quantity into the pre-training distribution network equipment to identify the depth convolution neural network after the depth image is processed by a bilateral filtering denoising algorithm.
As a preferable technical measure:
finishing the steps of equipment classification and position coordinate extraction in the network, if an unidentifiable model exists in the output, the field worker needs to shoot the depth image again and restart the process; and if all the models are successfully identified, outputting the quantity, the type and the central point coordinate parameters of the power distribution equipment.
As a preferable technical measure: the convolutional neural network is a feedforward convolutional neural network, and the CNNC structure of the convolutional neural network is divided into five layers:
performing Convolution (Convolution) on a first layer of input pictures to obtain a matrix (Feature Map) with a second layer depth of 3;
performing Pooling (Pooling) operation on the matrix of the second layer to obtain a matrix with the depth of the third layer being 3;
repeating the above operations to obtain a matrix with a fifth layer depth of 5, and finally expanding and connecting the 5 matrixes according to rows to form a vector, and transmitting the vector into a Fully Connected (full Connected) layer, wherein the Fully Connected layer is a BP convolutional neural network; each matrix can be viewed as neurons arranged in a matrix that is much the same as the neurons in a BP convolutional neural network.
As a preferable technical measure:
setting the number of layers for zero padding according to requirements; the Zero-filling layer is called Zero Padding and is a super-parameter which can be set, but the Zero-filling layer is adjusted according to the size and the step of the convolution kernel and the size of an input matrix so that the convolution kernel just slides to the edge;
generally, the input picture matrix and the subsequent convolution kernel, feature map matrix are both square matrices, where the input matrix size is w, the convolution kernel size is k, the stride is s, the number of zero padding layers is p, and then the calculation formula of the feature map size generated after convolution is:
as a preferable technical measure:
in order to extract more features, a plurality of convolution kernels are adopted to carry out convolution respectively, so that a plurality of feature maps can be obtained; sometimes, for a three-channel color picture, a set of matrices is input, and the convolution kernel is not layered any more and is changed to a corresponding depth.
Compared with the prior art, the invention has the following beneficial effects:
the method comprises the steps of collecting three-dimensional information and pictures of a power distribution room through laser depth sensor data, obtaining parameters required by drawing by using a convolutional neural network, and transmitting the parameters to a CAD module; finally, outputting a practically available power distribution room design drawing; the method can effectively shorten the design time of the business expansion and improve the design efficiency of the business expansion.
By applying the scheme of the invention, the existing high-voltage business expansion design process can be reduced from the original 5 working days to 0.5 working day, and the revolutionary improvement of the working efficiency is realized. The invention can organically integrate power supply service, power engineering design, power engineering construction, power equipment manufacturing and the like, shorten the design period, standardize the engineering standard, increase the supply and marketing, improve the user satisfaction degree, finally continuously promote the generation of enterprise value and form a business mode of virtuous circle development.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a flow chart of the operation of outputting the number of devices, the type of devices, and the location coordinates of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
On the contrary, the invention is intended to cover alternatives, modifications, equivalents and alternatives which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the present invention, certain specific details are set forth in order to provide a better understanding of the present invention. It will be apparent to one skilled in the art that the present invention may be practiced without these specific details.
As shown in fig. 1-2, a method for generating a cloud computing-based design of a power distribution room for a user,
which comprises the following steps:
firstly, establishing an interface program WSGI (Web service gateway initiative) which adopts a Web server and a Web application program defined for python programming voice and a depth calculation framework;
secondly, collecting three-dimensional information of the power distribution room by using a laser depth camera through staff on the site of the power distribution room;
thirdly, transmitting the three-dimensional reconstruction information and the laser depth photo of the equipment to a WSGI of a cloud server through a wireless network;
fourthly, transmitting data to a convolutional neural network inference module based on a GPU cluster through a WSGI (wireless sensor and antenna array) interface of the cloud end, acquiring parameters required by drawing, and transmitting the parameters to a python API (application programming interface) interface of a CAD (computer-aided design) module;
and fifthly, using a cloud accelerated computing architecture to accelerate the space computation of data collected by the field sensor, and finally outputting a standard drawing of field collection of the power distribution room.
The method comprises the steps of collecting three-dimensional information and pictures of a power distribution room through laser depth sensor data, obtaining parameters required by drawing by using a convolutional neural network, and transmitting the parameters to a CAD module; finally, outputting a practically available power distribution room design drawing; the method can effectively shorten the design time of the business expansion and improve the design efficiency of the business expansion.
By applying the scheme of the invention, the existing high-voltage business expansion design process can be reduced from the original 5 working days to 0.5 working day, and the revolutionary improvement of the working efficiency is realized. The invention can organically integrate power supply service, power engineering design, power engineering construction, power equipment manufacturing and the like, shorten the design period, standardize the engineering standard, increase the supply and marketing, improve the user satisfaction degree, finally continuously promote the generation of enterprise value and form a business mode of virtuous circle development.
The invention discloses a convolution neural network input source, which comprises the following specific embodiments:
the input data of the convolutional neural network is a distribution room three-dimensional reconstruction point cloud collected by a laser depth camera through a synchronous positioning and mapping technology SLAM, the point cloud applies a partial differential algorithm to obtain a point with large variation as a characteristic point cloud, and then a plurality of coordinate points are randomly extracted to serve as an input source of the convolutional neural network; the device image taken with the laser depth camera is used as another input to the convolutional neural network.
The convolutional neural network training simulation of the invention is a specific embodiment:
the training of the convolutional neural network is completed by a simulation system, and a three-dimensional engine for robot training, a gazebo, is adopted to randomly generate a plurality of power distribution rooms and equipment with different sizes; a virtual laser depth camera of the robot is used as sensor input in the system, the virtual laser depth sensor is installed on the body of the virtual robot, a robot SLAM technology is operated in an ROS (convolutional systematic) system to be used for autonomous navigation of the robot to collect space and equipment data, a data set for convolutional neural network training is automatically generated, and a pre-training network is generated through a GPU cluster; finally, design survey data, including the number of devices, the type of devices, and the coordinates of the devices relative to the three-dimensional reconstruction, are output through the pre-trained network.
The invention discloses a concrete embodiment of SLAM technology:
the SLAM technology aims to construct a surrounding environment map in real time according to virtual laser depth sensor data without any prior knowledge, and simultaneously presumes the self positioning according to the map, and a virtual robot carries a virtual laser depth sensor to move in an unknown environment;
for convenience, the movement of a continuous time is changed into discrete time t 1, … k, and at these times, the position of the virtual robot is represented by x, and then the position of each time is represented as x1, x2 … xk, which constitutes the track of the virtual robot;
in the aspect of a map, the map consists of a plurality of three-dimensional point coordinates, and at each moment, the virtual laser depth sensor can measure a part of the three-dimensional point coordinates to obtain observation data of the three-dimensional point coordinates;
the three-dimensional point coordinates are N in number and are represented by y1 and y2 … yn; the positioning problem x and the mapping problem y are solved by the motion measurement u and the virtual laser depth sensor reading z.
Another embodiment of the convolutional neural network input source of the present invention:
calling a laser ToF sensor, finishing three-dimensional point cloud collection of the power distribution room by staff on the site of the power distribution room, and judging the collected laser point cloud through a simple quantity;
the point cloud with the point number exceeding 50000 can be used as the input of a room key point coordinate extraction convolutional neural network, and no more than 50000 points require a worker to collect the three-dimensional point cloud again;
after the point cloud is input into the convolutional neural network, the coordinates of the key points of the basic shape of the room are inferred (reference) and vectorized.
The invention discloses a specific embodiment of invalid pixel point judgment, which comprises the following steps:
the method comprises the steps that field workers shoot depth images of existing high-voltage distribution equipment of a user by using the same laser ToF sensor, and the obtained laser depth images firstly pass through an invalid pixel point judgment process;
if the number of the invalid pixel points exceeds 15% of the total number of the pixel points, the worker is required to shoot the high-voltage distribution equipment again. And (3) inputting the depth image with invalid pixel points less than 15% of the total pixel quantity into the pre-training distribution network equipment to identify the depth convolution neural network after the depth image is processed by a bilateral filtering denoising algorithm.
Finishing the steps of equipment classification and position coordinate extraction in the network, if an unidentifiable model exists in the output, the field worker needs to shoot the depth image again and restart the process; and if all the models are successfully identified, outputting the quantity, the type and the central point coordinate parameters of the power distribution equipment.
One embodiment of the convolutional neural network architecture hierarchy of the present invention: the convolutional neural network is a feedforward convolutional neural network, and the CNNC structure of the convolutional neural network is divided into five layers:
performing Convolution (Convolution) on a first layer of input pictures to obtain a matrix (Feature Map) with a second layer depth of 3;
performing Pooling (Pooling) operation on the matrix of the second layer to obtain a matrix with the depth of the third layer being 3;
repeating the above operations to obtain a matrix with a fifth layer depth of 5, and finally expanding and connecting the 5 matrixes according to rows to form a vector, and transmitting the vector into a Fully Connected (full Connected) layer, wherein the Fully Connected layer is a BP convolutional neural network; each matrix can be viewed as neurons arranged in a matrix that is much the same as the neurons in a BP convolutional neural network.
Setting the number of layers for zero padding according to requirements; the Zero-filling layer is called Zero Padding and is a super-parameter which can be set, but the Zero-filling layer is adjusted according to the size and the step of the convolution kernel and the size of an input matrix so that the convolution kernel just slides to the edge;
generally, the input picture matrix and the subsequent convolution kernel, feature map matrix are both square matrices, where the input matrix size is w, the convolution kernel size is k, the stride is s, the number of zero padding layers is p, and then the calculation formula of the feature map size generated after convolution is:
in order to extract more features, a plurality of convolution kernels are adopted to carry out convolution respectively, so that a plurality of feature maps can be obtained; sometimes, for a three-channel color picture, a set of matrices is input, and the convolution kernel is not layered any more and is changed to a corresponding depth.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A user power distribution room design generation method based on cloud computing is characterized in that,
which comprises the following steps:
firstly, establishing an interface program WSGI (Web service gateway initiative) which adopts a Web server and a Web application program defined for python programming voice and a depth calculation framework;
secondly, collecting three-dimensional information of the power distribution room by using a laser depth camera through staff on the site of the power distribution room;
thirdly, transmitting the three-dimensional reconstruction information and the laser depth photo of the equipment to a WSGI of a cloud server through a wireless network;
fourthly, transmitting data to a convolutional neural network inference module based on a GPU cluster through a WSGI (wireless sensor and antenna array) interface of the cloud end, acquiring parameters required by drawing, and transmitting the parameters to a python API (application programming interface) interface of a CAD (computer-aided design) module;
and fifthly, using a cloud accelerated computing architecture to accelerate the space computation of data collected by the field sensor, and finally outputting a standard drawing of field collection of the power distribution room.
2. The cloud-computing-based method for generating a design of a power distribution room for a user according to claim 1,
the input data of the convolutional neural network is a distribution room three-dimensional reconstruction point cloud collected by a laser depth camera through a synchronous positioning and mapping technology SLAM, the point cloud applies a partial differential algorithm to obtain a point with large variation as a characteristic point cloud, and then a plurality of coordinate points are randomly extracted to serve as an input source of the convolutional neural network; the device image taken with the laser depth camera is used as another input to the convolutional neural network.
3. The cloud-computing-based method for generating a design of a power distribution room for a user according to claim 2,
the training of the convolutional neural network is completed by a simulation system, and a three-dimensional engine for robot training, a gazebo, is adopted to randomly generate a plurality of power distribution rooms and equipment with different sizes; the virtual laser depth camera of the robot is used as sensor input in the system, the virtual laser depth sensor is installed on the body of the virtual robot, the robot SLAM technology is operated in the ROS system and used for autonomous navigation of the robot to collect space and equipment data, a data set for convolutional neural network training is automatically generated, and a pre-training network is generated through a GPU cluster; finally, design survey data, including the number of devices, the type of devices, and the coordinates of the devices relative to the three-dimensional reconstruction, are output through the pre-trained network.
4. The cloud-computing-based method for generating a design of a power distribution room for a user according to claim 3,
the SLAM technology aims to construct a surrounding environment map in real time according to virtual laser depth sensor data without any prior knowledge, and simultaneously presumes the self positioning according to the map, and a virtual robot carries a virtual laser depth sensor to move in an unknown environment;
changing the motion of a continuous time into discrete time t as 1, … k, and at these time, using x to represent the self position of the virtual robot, then the position of each time is marked as x1, x2 … xk, which constitutes the track of the virtual robot;
in the aspect of a map, the map consists of a plurality of three-dimensional point coordinates, and at each moment, the virtual laser depth sensor can measure a part of the three-dimensional point coordinates to obtain observation data of the three-dimensional point coordinates;
the three-dimensional point coordinates are N in number and are represented by y1 and y2 … yn; the positioning problem x and the mapping problem y are solved by the motion measurement u and the virtual laser depth sensor reading z.
5. The cloud-computing-based method for generating a design of a power distribution room for a user according to claim 1,
calling a laser ToF sensor, finishing three-dimensional point cloud collection of the power distribution room by staff on the site of the power distribution room, and judging the collected laser point cloud through a simple quantity;
the point cloud with the point number exceeding 50000 can be used as the input of a room key point coordinate extraction convolutional neural network, and no more than 50000 points require a worker to collect the three-dimensional point cloud again;
after the point cloud is input into the convolutional neural network, the coordinates of the key points of the basic shape of the room are inferred (reference) and vectorized.
6. The cloud-computing-based method for generating a design of a power distribution room for a user according to claim 5,
the method comprises the steps that field workers shoot depth images of existing high-voltage distribution equipment of a user by using the same laser ToF sensor, and the obtained laser depth images firstly pass through an invalid pixel point judgment process;
if the number of the invalid pixel points exceeds 15% of the total number of the pixel points, the worker is required to shoot the high-voltage distribution equipment again. And (3) inputting the depth image with invalid pixel points less than 15% of the total pixel quantity into the pre-training distribution network equipment to identify the depth convolution neural network after the depth image is processed by a bilateral filtering denoising algorithm.
7. The cloud-computing-based method for generating a design of a user distribution room as claimed in claim 6,
finishing the steps of equipment classification and position coordinate extraction in the network, if an unidentifiable model exists in the output, the field worker needs to shoot the depth image again and restart the process; and if all the models are successfully identified, outputting the quantity, the type and the central point coordinate parameters of the power distribution equipment.
8. The cloud-computing-based user power distribution room design generation method as claimed in any one of claims 1 to 7, wherein the convolutional neural network is a feedforward type convolutional neural network, and a CNNC structure of the convolutional neural network is divided into five layers:
the first layer inputs pictures, and convolution operation is carried out to obtain a matrix with the depth of the second layer being 3;
performing pooling operation on the matrix of the second layer to obtain a matrix with the depth of the third layer being 3;
repeating the above operations to obtain a matrix with the fifth layer depth of 5, and finally expanding and connecting the 5 matrixes according to rows to form vectors, and transmitting the vectors into a full connection layer which is a BP convolutional neural network; each matrix can be viewed as neurons arranged in a matrix that is much the same as the neurons in a BP convolutional neural network.
9. The cloud-computing-based method for generating a design of a power distribution room for a user according to claim 8,
setting the number of layers for zero padding according to requirements; the Zero-filling layer is called Zero Padding and is a super-parameter which can be set, but the Zero-filling layer is adjusted according to the size and the step of the convolution kernel and the size of an input matrix so that the convolution kernel just slides to the edge;
generally, the input picture matrix and the subsequent convolution kernel, feature map matrix are both square matrices, where the input matrix size is w, the convolution kernel size is k, the stride is s, the number of zero padding layers is p, and then the calculation formula of the feature map size generated after convolution is:
10. the cloud-computing-based method for generating a design of a power distribution room for a user according to claim 9,
in order to extract more features, a plurality of convolution kernels are adopted to carry out convolution respectively, so that a plurality of feature maps can be obtained; for a three-channel color picture, a set of matrices is input, and the convolution kernel is not layered any more and is changed to a corresponding depth.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911341997.6A CN111192363B (en) | 2019-12-23 | 2019-12-23 | User power distribution room design generation method based on cloud computing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911341997.6A CN111192363B (en) | 2019-12-23 | 2019-12-23 | User power distribution room design generation method based on cloud computing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111192363A true CN111192363A (en) | 2020-05-22 |
CN111192363B CN111192363B (en) | 2023-09-29 |
Family
ID=70707457
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911341997.6A Active CN111192363B (en) | 2019-12-23 | 2019-12-23 | User power distribution room design generation method based on cloud computing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111192363B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111832542A (en) * | 2020-08-15 | 2020-10-27 | 武汉易思达科技有限公司 | Three-eye visual identification and positioning method and device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103400415A (en) * | 2013-08-07 | 2013-11-20 | 赞奇科技发展有限公司 | Implementation method and system for automatic generation and on-line interaction of three-dimensional digital model with planar space structure |
CN105027190A (en) * | 2013-01-03 | 2015-11-04 | 美达公司 | Extramissive spatial imaging digital eye glass for virtual or augmediated vision |
CN108171796A (en) * | 2017-12-25 | 2018-06-15 | 燕山大学 | A kind of inspection machine human visual system and control method based on three-dimensional point cloud |
CN108365994A (en) * | 2018-03-13 | 2018-08-03 | 山东超越数控电子股份有限公司 | A kind of cloud security management platform for cloud computing security and unity management |
CN108921893A (en) * | 2018-04-24 | 2018-11-30 | 华南理工大学 | A kind of image cloud computing method and system based on online deep learning SLAM |
CN109801358A (en) * | 2018-12-06 | 2019-05-24 | 宁波市电力设计院有限公司 | A kind of substation's three-dimensional investigation method scanning and put cloud visual fusion based on SLAM |
-
2019
- 2019-12-23 CN CN201911341997.6A patent/CN111192363B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105027190A (en) * | 2013-01-03 | 2015-11-04 | 美达公司 | Extramissive spatial imaging digital eye glass for virtual or augmediated vision |
CN103400415A (en) * | 2013-08-07 | 2013-11-20 | 赞奇科技发展有限公司 | Implementation method and system for automatic generation and on-line interaction of three-dimensional digital model with planar space structure |
CN108171796A (en) * | 2017-12-25 | 2018-06-15 | 燕山大学 | A kind of inspection machine human visual system and control method based on three-dimensional point cloud |
CN108365994A (en) * | 2018-03-13 | 2018-08-03 | 山东超越数控电子股份有限公司 | A kind of cloud security management platform for cloud computing security and unity management |
CN108921893A (en) * | 2018-04-24 | 2018-11-30 | 华南理工大学 | A kind of image cloud computing method and system based on online deep learning SLAM |
CN109801358A (en) * | 2018-12-06 | 2019-05-24 | 宁波市电力设计院有限公司 | A kind of substation's three-dimensional investigation method scanning and put cloud visual fusion based on SLAM |
Non-Patent Citations (1)
Title |
---|
王权等: "基于ROS与Kinect的移动机器人同时定位与地图构建" * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111832542A (en) * | 2020-08-15 | 2020-10-27 | 武汉易思达科技有限公司 | Three-eye visual identification and positioning method and device |
CN111832542B (en) * | 2020-08-15 | 2024-04-16 | 武汉易思达科技有限公司 | Tri-vision identifying and positioning device |
Also Published As
Publication number | Publication date |
---|---|
CN111192363B (en) | 2023-09-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110288695A (en) | Single-frame images threedimensional model method of surface reconstruction based on deep learning | |
CN109978767A (en) | The ground laser SLAM drawing method based on multirobot collaboration | |
CN111028335B (en) | Point cloud data block surface patch reconstruction method based on deep learning | |
CN108154104A (en) | A kind of estimation method of human posture based on depth image super-pixel union feature | |
CN112862090A (en) | Air temperature forecasting method based on deep space-time neural network | |
CN111709397A (en) | Unmanned aerial vehicle variable-size target detection method based on multi-head self-attention mechanism | |
CN114693064B (en) | Building group scheme generation performance evaluation method | |
CN103337085A (en) | Efficient portrait face distortion method | |
CN109583377A (en) | A kind of control method, device and host computer that pipeline model is rebuild | |
CN111104702A (en) | Method for realizing visual simulation of embankment project based on UE4 | |
CN115578393A (en) | Key point detection method, key point training method, key point detection device, key point training device, key point detection equipment, key point detection medium and key point detection medium | |
CN112148820A (en) | Underwater terrain data identification and service method and system based on deep learning | |
CN111192363B (en) | User power distribution room design generation method based on cloud computing | |
CN111047703B (en) | User high-voltage distribution equipment identification and space reconstruction method | |
CN116958478B (en) | City building model programming generation method, device, equipment and storage medium | |
CN112017159A (en) | Ground target reality simulation method in remote sensing scene | |
CN112488413A (en) | AWA-DRCN-based population spatialization method | |
CN115849202B (en) | Intelligent crane operation target identification method based on digital twin technology | |
CN105321205B (en) | A kind of parameterized human body model method for reconstructing based on sparse key point | |
CN104463934A (en) | Automatic generation method for point set model animation driven by mass point-spring system | |
CN116051699A (en) | Dynamic capture data processing method, device, equipment and storage medium | |
Jia | Daylight prediction using Gan: General workflow, tool development and case study on Manhattan, New York | |
CN113192204B (en) | Three-dimensional reconstruction method for building in single inclined remote sensing image | |
CN110689616B (en) | Water delivery channel parametric modeling method based on three-dimensional digital earth | |
CN112837420A (en) | Method and system for completing shape of terracotta warriors point cloud based on multi-scale and folding structure |
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 |