CN113346966A - Channel feedback method for unmanned aerial vehicle inspection communication subsystem of smart power grid - Google Patents
Channel feedback method for unmanned aerial vehicle inspection communication subsystem of smart power grid Download PDFInfo
- Publication number
- CN113346966A CN113346966A CN202110584019.5A CN202110584019A CN113346966A CN 113346966 A CN113346966 A CN 113346966A CN 202110584019 A CN202110584019 A CN 202110584019A CN 113346966 A CN113346966 A CN 113346966A
- Authority
- CN
- China
- Prior art keywords
- aerial vehicle
- unmanned aerial
- vector
- channel
- receiving base
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000004891 communication Methods 0.000 title claims abstract description 18
- 238000007689 inspection Methods 0.000 title claims abstract description 16
- 230000006837 decompression Effects 0.000 claims description 12
- 230000006835 compression Effects 0.000 claims description 11
- 238000007906 compression Methods 0.000 claims description 11
- 238000011084 recovery Methods 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 abstract description 14
- 230000006870 function Effects 0.000 description 13
- 230000007246 mechanism Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000012512 characterization method Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 230000007787 long-term memory Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/391—Modelling the propagation channel
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/183—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a single remote source
- H04N7/185—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a single remote source from a mobile camera, e.g. for remote control
Abstract
The invention discloses a smart power grid unmanned aerial vehicle routing inspection communication subsystem channel feedback method, which relates to a configuration comprising an unmanned aerial vehicle and a plurality of electric power towers, wherein each electric power tower is provided with a receiving base station for receiving data transmitted by the unmanned aerial vehicle; the method comprises the following steps: step 1: the unmanned aerial vehicle transmits the detected video pictures to a plurality of receiving base stations in a segmented manner in real time; step 2: and summarizing and transmitting the data received by the plurality of receiving base stations to a ground master control station. A system model of the unmanned aerial vehicle terminal and the base station under the scene is established by establishing the return access points in a segmented mode, and the problems of real-time video picture transmission and real-time channel coefficient feedback are solved.
Description
Technical Field
The invention relates to the technical field of unmanned aerial vehicle inspection, in particular to a channel feedback method of an unmanned aerial vehicle inspection communication subsystem of a smart power grid.
Background
In the prior art, when the unmanned aerial vehicle patrols and examines, the unmanned aerial vehicle changes in real time with the channel environment and the channel transmission coefficient of the base station, and the channel transmission coefficient after the change needs to be estimated and tracked in real time and fed back to the base station. The complete channel coefficient feedback consumes a large amount of link overhead, so vector quantization or codebook-based methods are usually adopted to reduce the feedback overhead.
The prior art has the following defects: first, a large number of antennas are used at a base station end in a massive MIMO system, so that the codebook design complexity and the corresponding feedback amount are significantly increased, and therefore, the conventional feedback scheme is not desirable in a massive MIMO-mm-wave unmanned aerial vehicle communication scenario. Secondly, the feedback problem of unmanned aerial vehicle time varying channel can not be solved to current feedback scheme.
Disclosure of Invention
The invention aims to provide a channel feedback method of an unmanned aerial vehicle inspection communication subsystem of a smart power grid, and the channel feedback method is used for solving the problem of overlarge feedback overhead in a time-varying channel.
The purpose of the invention is realized as follows: a smart power grid unmanned aerial vehicle routing inspection communication subsystem channel feedback method relates to a configuration comprising an unmanned aerial vehicle and a plurality of electric power towers, wherein each electric power tower is provided with a receiving base station for receiving data sent by the unmanned aerial vehicle;
the method comprises the following steps:
step 1: the unmanned aerial vehicle transmits the detected video pictures to a plurality of receiving base stations in a segmented manner in real time;
step 2: and summarizing and transmitting the data received by the plurality of receiving base stations to a ground master control station.
Furthermore, an encoder is configured at the unmanned aerial vehicle end, and the encoder is formed by connecting a feature extraction module and a feature compression module in series;
the feature extraction module is mainly composed of a convolution layer of convolution kernel 3 × 3, a regularization layer of BatchNorm and a Leaky Relu activation function;
in the feature compression module, the extracted feature information matrix is converted into a vector, and the vector is input into the FCN and LSTM-Attention modules which are connected in parallel to generate a codeword vector s.
Further, each receiving base station side recovers the channel state information of the code word vector s from the unmanned aerial vehicle side through a decoder of the receiving base station side.
Furthermore, the decoder at the receiving base station end consists of a characteristic decompression module and a channel recovery module;
the decompression module decompresses the vector with the size of M x 1 into the vector with the size of N x 1, after decompression, the output value is the initialized estimation value of the real part and the imaginary part of the channel matrix, and the decompressed vector is input into the channel recovery module to recover the channel state information;
and inputting the vector into two RefineNet units at a channel recovery module, wherein the RefineNet units comprise 3 Conv3D layers of 3 × 3, are laminated one by one, and respectively output 8, 16 and 2 characteristic diagrams, and meanwhile, a regularization layer of Batch-norm and a LeakyRelu activation function are connected behind each convolution layer, and feature information is fused in the units and activated through the LeakyRelu function, so that the output value of the Refine unit is finally obtained.
The invention has the beneficial effects that:
due to the fact that the transmission line is too long, the data quantity of stored routing inspection video pictures of the unmanned aerial vehicle is too large, the detected video pictures need to be transmitted to the ground receiving base station in real time, a system model of the unmanned aerial vehicle terminal and the base station under the scene is built through the segmented building of the return access points, and the problems of real-time transmission of the video pictures and real-time feedback of channel coefficients are solved.
Drawings
Fig. 1 is a schematic diagram of a communication scenario of an unmanned aerial vehicle according to the present invention.
Fig. 2 is a schematic diagram of an unmanned-end encoder.
Fig. 3 is a schematic diagram of a decoder at the base station side.
Detailed Description
The invention will be further described with reference to the accompanying figures 1-3 and specific examples.
As shown in fig. 1 to 3, this embodiment provides a smart grid unmanned aerial vehicle inspection communication subsystem channel feedback method, which includes:
firstly, configuring an unmanned aerial vehicle end (encoder): the encoder is composed of a feature extraction module and a feature compression module which are connected in series, the feature extraction module is mainly composed of a convolution layer of convolution kernel 3 x 3, a regularization layer of BatchNorm and a Leaky Relu activation function, in the feature compression module, an extracted feature information matrix is converted into a vector, the vector is input into an FCN and an LSTM-Attention module which are connected in parallel to generate a code word vector s, wherein the FCN can accelerate convergence speed and effectively relieve the problem of gradient disappearance, the LSTM is used for learning time correlation characteristics among channel matrixes, and an Attention mechanism enables a neural network to have an automatic weighting function, so that the characterization capability of feature information is further improved.
Secondly, configuring a base station end (decoder): the method comprises the steps of recovering channel state information from a code word vector s from an unmanned machine end through a decoder, wherein the decoder consists of a characteristic decompression module and a channel recovery module, the characteristic decompression module is similar to the encoder, the difference is that the decompression module decompresses the vector with the size of M1 into the vector with the size of N1, the decompressed vector outputs initialized estimation values of a real part and an imaginary part of a channel matrix, the decompressed vector is input into the channel recovery module to recover the channel state information, the vector is input into two Refinelet units in the channel recovery module, the Refinelet units comprise 3 Conv3D layers with 3 x 3, the Conv3D layers are laminated layer by layer and output 8, 16 and 2 characteristic diagrams respectively, and each convolution layer is connected with a Batch-norm regular layer, a LeyRelu activation function, and the characteristic information is fused in the units, and activating through a LeakyRelu function to finally obtain an output value of the Refine unit.
Thirdly, pretreatment: before transmission, the CSI matrix requires two pre-processes, H is first sparse in the angular delay domain after a 2D Discrete Fourier Transform (DFT) operation, then in the delay domain, most elements in H are zero (except the first few non-zero columns), since the time delay between multipath arrivals around the straight path is within a finite time period, so the first Nc non-zero column can be retained while the remaining non-zero columns are deleted, and a new CSI matrix of size Nt Nc is represented as H, which can perform preliminary compression on the channel matrix coefficients.
Training: the network trains and optimizes the two channel feedback network frames in an end-to-end mode in a cyclic iteration mode, the optimization aims to minimize a loss function, and finally obtains an optimal network kernel function and an optimal network deviation through a gradient descent method so as to obtain an optimized neural network.
As shown in fig. 1, according to the actual unmanned aerial vehicle power transmission line inspection case, it is considered that the transmitting power of the unmanned aerial vehicle communication equipment is limited, every 5km distance (the inspection speed of a multi-rotor unmanned aerial vehicle is about 30km/h), a receiving base station is built on each power tower, the unmanned aerial vehicle transmits high-definition video pictures collected by a high-definition camera or an infrared camera to the receiving base station in real time, the high-definition video pictures are transmitted to a plurality of receiving base stations in a segmented mode, the base stations transmit the high-definition video pictures to a ground master control station (Mesh wireless gateway can be collected during data transmission) through communication lines carried on the power towers, and the problems that the power transmission line is too long and the transmission and the storage data amount are too large are solved.
In this embodiment, the following are set:
A) unmanned aerial vehicle patrols and examines communication subsystem, the effect is: because transmission line overlength, unmanned aerial vehicle storage patrols and examines video picture data volume too big, need transmit the video picture that detects in real time and accept the basic station on ground. Under a large-scale MIMO millimeter wave unmanned aerial vehicle communication scene, a system model of an unmanned aerial vehicle terminal and a base station under the scene is established by establishing a return access point in a segmented manner, so that the problems of video picture real-time transmission and channel coefficient real-time feedback are solved;
B) the deep learning automatic encoder framework has the functions of: for the problem of channel coefficient feedback of an unmanned aerial vehicle communication subsystem, an automatic encoder in deep learning is adopted to form a feedback network framework, wherein the functions of the network comprise feature extraction and compression in the encoder, decompression and channel reconstruction of the encoder, and the automatic encoder and a decoder are formed by network structures such as a convolutional neural network, a full-connection network, long-term and short-term memory, an attention mechanism and the like. The long-term and short-term memory network can make full use of the time-varying channel characteristics in the unmanned aerial vehicle inspection process to improve the compression reconstruction performance, and meanwhile, the attention mechanism is added to enable the neural network to have the automatic weighting function, so that the characterization capability of the characteristic information is further improved.
The embodiment has the following advantages:
1. due to the fact that the transmission line is too long, the data quantity of stored polling video pictures of the unmanned aerial vehicle is too large, the detected video pictures need to be transmitted to a ground receiving base station in real time, a system model of the unmanned aerial vehicle terminal and the base station under the scene is built through subsection building of a return access point, and the problems of real-time transmission of the video pictures and real-time feedback of channel coefficients are solved;
2. a novel and effective CSI sensing and recovery mechanism called UAVCsNet is proposed, which utilizes the storage characteristics of RNN in the feature extraction, compression and decompression modules respectively, and furthermore, we adopt deep separable convolution in the feature recovery to reduce the size of the model and enhance the mutual information between channels. Thereby improving the feedback reconstruction performance;
3. a long-short term memory-attention focusing mechanism network is added in a compression and decompression module of the network, the problem of the time-varying channel of the unmanned aerial vehicle is solved, the characteristics of the unmanned aerial vehicle are fully utilized, the neural network has an automatic weighting function, and the characterization capability of characteristic information is further improved.
While the preferred embodiments of the present invention have been described, those skilled in the art will appreciate that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (4)
1. A smart grid unmanned aerial vehicle routing inspection communication subsystem channel feedback method is characterized in that the configuration related to the method comprises an unmanned aerial vehicle and a plurality of electric power towers, wherein each electric power tower is provided with a receiving base station for receiving data sent by the unmanned aerial vehicle;
the method comprises the following steps:
step 1: the unmanned aerial vehicle transmits the detected video pictures to a plurality of receiving base stations in a segmented manner in real time;
step 2: and summarizing and transmitting the data received by the plurality of receiving base stations to a ground master control station.
2. The smart grid unmanned aerial vehicle inspection communication subsystem channel feedback method according to claim 1, wherein: an encoder is configured at the unmanned aerial vehicle end, and the encoder is formed by connecting a feature extraction module and a feature compression module in series;
the feature extraction module is mainly composed of a convolution layer of convolution kernel 3 × 3, a regularization layer of BatchNorm and a Leaky Relu activation function;
in the feature compression module, the extracted feature information matrix is converted into a vector, and the vector is input into the FCN and LSTM-Attention modules which are connected in parallel to generate a codeword vector s.
3. The smart grid unmanned aerial vehicle inspection communication subsystem channel feedback method according to claim 1, wherein: and each receiving base station side recovers the channel state information of the code word vector s from the unmanned aerial vehicle side through a decoder of the receiving base station side.
4. The smart grid unmanned aerial vehicle inspection communication subsystem channel feedback method according to claim 3, wherein the method comprises the following steps: the decoder of the receiving base station end consists of a characteristic decompression module and a channel recovery module;
the decompression module decompresses the vector with the size of M x 1 into the vector with the size of N x 1, after decompression, the output value is the initialized estimation value of the real part and the imaginary part of the channel matrix, and the decompressed vector is input into the channel recovery module to recover the channel state information;
and inputting the vector into two RefineNet units at a channel recovery module, wherein the RefineNet units comprise 3 Conv3D layers of 3 × 3, are laminated one by one, and respectively output 8, 16 and 2 characteristic diagrams, and meanwhile, a regularization layer of Batch-norm and a LeakyRelu activation function are connected behind each convolution layer, and feature information is fused in the units and activated through the LeakyRelu function, so that the output value of the Refine unit is finally obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110584019.5A CN113346966A (en) | 2021-05-27 | 2021-05-27 | Channel feedback method for unmanned aerial vehicle inspection communication subsystem of smart power grid |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110584019.5A CN113346966A (en) | 2021-05-27 | 2021-05-27 | Channel feedback method for unmanned aerial vehicle inspection communication subsystem of smart power grid |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113346966A true CN113346966A (en) | 2021-09-03 |
Family
ID=77471839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110584019.5A Pending CN113346966A (en) | 2021-05-27 | 2021-05-27 | Channel feedback method for unmanned aerial vehicle inspection communication subsystem of smart power grid |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113346966A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114095065A (en) * | 2021-09-23 | 2022-02-25 | 上海电机学院 | Hybrid beam forming method for intelligent unmanned aerial vehicle inspection based on deep learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104902243A (en) * | 2015-06-09 | 2015-09-09 | 四川汇源光通信有限公司 | Unmanned plane line patrolling high-definition video real-time return system based on optical fiber ring network, and return method |
| CN109474316A (en) * | 2018-11-22 | 2019-03-15 | 东南大学 | A kind of channel information compression feedback method based on deep-cycle neural network |
| CN110912598A (en) * | 2019-11-22 | 2020-03-24 | 中原工学院 | Large-scale MIMO system CSI feedback method based on long-time attention mechanism |
| CN112511197A (en) * | 2020-12-01 | 2021-03-16 | 南京工业大学 | Unmanned aerial vehicle auxiliary elastic video multicast method based on deep reinforcement learning |
| CN112533225A (en) * | 2020-12-29 | 2021-03-19 | 上海瀚讯信息技术股份有限公司 | Full-duplex wireless network based on vortex wave beams and node pairing method thereof |
-
2021
- 2021-05-27 CN CN202110584019.5A patent/CN113346966A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104902243A (en) * | 2015-06-09 | 2015-09-09 | 四川汇源光通信有限公司 | Unmanned plane line patrolling high-definition video real-time return system based on optical fiber ring network, and return method |
| CN109474316A (en) * | 2018-11-22 | 2019-03-15 | 东南大学 | A kind of channel information compression feedback method based on deep-cycle neural network |
| CN110912598A (en) * | 2019-11-22 | 2020-03-24 | 中原工学院 | Large-scale MIMO system CSI feedback method based on long-time attention mechanism |
| CN112511197A (en) * | 2020-12-01 | 2021-03-16 | 南京工业大学 | Unmanned aerial vehicle auxiliary elastic video multicast method based on deep reinforcement learning |
| CN112533225A (en) * | 2020-12-29 | 2021-03-19 | 上海瀚讯信息技术股份有限公司 | Full-duplex wireless network based on vortex wave beams and node pairing method thereof |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114095065A (en) * | 2021-09-23 | 2022-02-25 | 上海电机学院 | Hybrid beam forming method for intelligent unmanned aerial vehicle inspection based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Qin et al. | Federated learning and wireless communications | |
| CN110417456B (en) | Information transmission method based on unmanned aerial vehicle | |
| CN112784718B (en) | Insulator state identification method based on edge calculation and deep learning | |
| CN109687897B (en) | Superposition CSI feedback method based on deep learning large-scale MIMO system | |
| CN111555781B (en) | Large-scale MIMO channel state information compression and reconstruction method based on deep learning attention mechanism | |
| CN107205154A (en) | A kind of radio multimedia sensor network compression of images acquisition method based on matrix completion | |
| CN113194548B (en) | Intelligent super-surface-assisted large-scale passive random access method | |
| CN113346966A (en) | Channel feedback method for unmanned aerial vehicle inspection communication subsystem of smart power grid | |
| Luo et al. | Multi-modal and multi-user semantic communications for channel-level information fusion | |
| CN116192209A (en) | Gradient uploading method for air computing federal learning under MIMO channel | |
| CN112929304A (en) | Wireless channel acquisition method assisted by environment knowledge base | |
| CN109672885A (en) | A kind of video image encoding and decoding method for mine intelligent monitoring | |
| CN116128070B (en) | Federal learning method based on wireless air calculation and multi-bit quantization compressed sensing | |
| CN109308517B (en) | Binary device, method and application for binary neural network | |
| Huang et al. | Research on channel feedback algorithm in UAV inspection communication subsystem of smart grid | |
| CN112261685B (en) | MIMO signal detection method based on convolutional neural network | |
| WO2023115254A1 (en) | Data processing method and device | |
| WO2023019585A1 (en) | Precoding model training method and apparatus, and precoding method and apparatus | |
| CN115333900B (en) | Unmanned aerial vehicle collaborative channel estimation and CSI feedback method based on deep learning | |
| WO2023231881A1 (en) | Model application method and apparatus | |
| CN115378483A (en) | Superimposed CSI feedback method based on LoS perception UAV millimeter wave system | |
| CN113660015B (en) | Online wireless channel acquisition optimization method under assistance of environment knowledge base | |
| WO2023070675A1 (en) | Data processing method and apparatus | |
| CN113422625B (en) | Topological interference management method and device based on sparse low-rank decomposition | |
| CN113347657B (en) | Method for predicting safety capacity performance of mobile communication system |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210903 |