CN115908897A - High-resolution remote sensing image-based intelligent identification method for high-speed railway power supply facilities - Google Patents
High-resolution remote sensing image-based intelligent identification method for high-speed railway power supply facilities Download PDFInfo
- Publication number
- CN115908897A CN115908897A CN202211339322.XA CN202211339322A CN115908897A CN 115908897 A CN115908897 A CN 115908897A CN 202211339322 A CN202211339322 A CN 202211339322A CN 115908897 A CN115908897 A CN 115908897A
- Authority
- CN
- China
- Prior art keywords
- power supply
- remote sensing
- sensing image
- identification
- speed railway
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of remote sensing image processing, in particular to an intelligent identification method for a high-speed railway power supply facility based on a high-resolution remote sensing image, which comprises the following steps: step S1: selecting a high-resolution remote sensing image, and interpreting and drawing samples in the area around the high-speed train station; step S2: setting initial parameters, calculating a loss function by using a sample with a boundary frame of the real position of the power supply network rod, and iterating and updating the parameter weight W and the offset value b in the process of participating in back propagation so as to finish the training of the model; and step S3: calculating by adopting a way of road network searching; and step S4: and randomly selecting some station images, putting the station images into a trained model for power supply network rod identification, and checking an identification result. The invention can realize automatic identification and extraction of the high-speed railway power supply facility, greatly improve the identification efficiency and ensure the stability of target detection.
Description
Technical Field
The invention relates to the technical field of remote sensing image processing, in particular to an intelligent identification method for a high-speed railway power supply facility based on a high-resolution remote sensing image.
Background
The power source of the high-speed train is electric energy, and the pantograph and catenary system is a power supply mode of the core of the high-speed train, so that the power supply facility of the high-speed railway mainly refers to pantograph and catenary contact net rods in the pantograph and catenary system. The railway line of the high-speed railway is often erected in a natural environment without human smoke, and the railway line runs with water and the situation that forests are brought in and out is more normal, so that in the detection of the condition of a railway power supply facility, particularly in the problem that whether the natural environment and the artificial environment near a bow net are dangerous to the bow net or even damage the bow net system, hands are organized to enter the old forest in the deep mountains along the high-speed railway line, and mountain and water are involved, the manual inspection of the railway line network with large size is not only labor and money, but also has no efficiency and can not be detected in real time, and on the premise that the manual interpretation-based method is continuously used for the identification and target extraction of large-size areas, and the efficient and real-time inspection and prevention are also not desirable.
The traditional identification classification is interpreted depending on manual experience, so that the efficiency is low, the timeliness is not available, the efficiency can be realized only in the aspect of solving the target detection of a small-range area, once the requirement scale is expanded to a provincial area, the efficiency is not high, the manpower is wasted, the inefficiency is natural, the timeliness is not available, and the event threatening the high-speed bow net power supply facility, which is required or occurring, cannot be detected by monitoring the research area at every moment.
Disclosure of Invention
The invention provides an intelligent identification method of a high-speed railway power supply facility based on a high-resolution remote sensing image, which can efficiently and accurately identify the high-speed railway power supply facility through a small amount of samples and limited calculation support.
The invention discloses an intelligent identification method of a high-speed railway power supply facility based on a high-resolution remote sensing image, which comprises the following steps:
step S1: selecting a high-resolution remote sensing image, and interpreting and drawing samples in the area around the high-speed train station;
step S2: setting initial parameters, calculating a loss function by using a sample with a boundary frame of the real position of the power supply network rod, and iterating and updating the parameter weight W and the offset value b in the process of participating in back propagation so as to finish the training of the model;
and step S3: calculating by adopting a way of road network searching;
and step S4: and randomly selecting some station images, putting the station images into a trained model for power supply network rod identification, and checking an identification result.
Preferably, the specific steps of step S1 are:
step S11: according to the high-resolution satellite remote sensing image, the position of the power supply network rod is sketched;
step S12: generating a minimum envelope rectangle for the position of a power supply network trunk by adopting a GIS tool;
step S13: and cutting the original image according to the drawing area, and finally respectively storing the shape file of the surface element, the KML file of the envelope rectangle and the TIFF file of the cut image.
Preferably, the specific steps of step S2 are:
step S21: placing the original image into a convolutional network FastRCNN to extract characteristic information, wherein the loss function of the FastRCNN is as follows:
L(p,u,t u ,v)＝L cls (p,u)+λ[u≥1]L loc (t u ,v) (1)
where p is the probability distribution predicted by the classifier, u corresponds to the target true class label, t u Regression parameters of the corresponding category u predicted by the corresponding boundary box regressor, and v corresponding to the boundary box regression parameters of the real target;
step S22: the feature image is used as a template and put into an RPN network to screen a boundary frame, then the boundary frame is further regressed to obtain a more accurate position, and coordinate information of the boundary frame is mapped onto a feature map, wherein the RPN network adopts the following loss function, and adopts individual discrimination as an agent task to construct a feature learning model in an automatic supervision mode, and parameters of the model are pre-trained:
wherein i refers to the index of an anchor in a mini-batch, p i An estimated probability representing whether the anchor point i is an object; if the anchor point is a positive sample, true mark
1, negative example 0; t is t i Vector of 4 parameterized coordinates representing a prediction bounding box, </or >>
Representing a real bounding box; loss of classification L cls Covering both classesLoss of logarithm; use { (R) } for regression loss>
Wherein R represents a robust smooth L1 loss; />
Represents the regression loss for sample anchor points only; the outputs of the cls and reg branches are respectively composed of { p i And t i Is formed by the following steps;
s23, putting the feature map with the bounding box into an ROI Pooling layer to unify the dimensions of the bounding box and outputting a feature vector with a fixed size;
and S24, calculating the category in the fully-connected layer by using a Softmax classifier and performing regression of the bounding box again to obtain more accurate positioning.
Preferably, the specific steps of step S3 are:
step S31: performing initialization prediction by adopting the model obtained by S22 training according to the initially determined range of S11;
step S32: when a station appears in a prediction result, establishing a root node; and further, area grid searching is carried out near the station, sub-nodes are established when the railway exists nearby, all railway lines are further traversed through the area grid searching, the sub-nodes are returned when the power supply facility searching is completed, and the whole area power supply facility identification and extraction are sequentially completed by continuing to predict along the advancing direction of the railway.
Preferably, the specific steps of step S4 are:
step S41: randomly selecting some station images as test data to be placed in the model for power supply network rod identification, and testing the confidence coefficient of the identification result;
step S42: and (3) considering the influence of different landforms on the power supply facility identification monitoring, inputting the high-speed railway images under different landforms into the model, and checking the robustness of the model under the condition of coping with different influences.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. according to the invention, the satellite image is adopted to obtain the sample to train the fast R-CNN deep learning model, so that automatic identification and extraction of the high-speed railway power supply facility can be realized, the identification efficiency is greatly improved, and the stability of target detection is ensured;
2. according to the method, a high-speed railway power supply facility identification scheme based on region search is constructed, and the scheme enables a model to be converted from a region traversal mode to a region search mode by establishing a root node-child node mode, so that the computational power loss is effectively reduced, and the prediction speed is greatly accelerated;
3. the recognition results under different landforms show that the model has very good effect and little difference even if certain landform influence exists.
Drawings
FIG. 1 is a flow chart of a method for intelligently identifying a power supply facility of a high-speed railway based on a high-resolution remote sensing image in an embodiment;
FIG. 2 is an example of generating a minimum envelope rectangle containing the grid bar positions in an embodiment;
fig. 3 is a grid diagram of stations and power grid poles outlined in the example.
Detailed Description
For a further understanding of the invention, reference should be made to the following detailed description taken in conjunction with the accompanying drawings and examples. It is to be understood that the examples are illustrative of the invention and not limiting.
Examples
As shown in fig. 1, the embodiment provides an intelligent identification method for a high-speed railway power supply facility based on a high-resolution remote sensing image, which includes the following steps:
step S1: and selecting a high-resolution remote sensing image, and interpreting and drawing samples in an area around the high-speed train station by about 1000 meters.
Step S2: setting initial parameters, calculating a loss function by using a sample with a bounding box of the real position of the power supply network rod, and iterating and updating the parameter weight W and the offset value b in the process of participating in back propagation, thereby completing the training of the model.
And step S3: considering that only a small part of the high-resolution remote sensing image contains high-speed railway power supply facilities, and the rest can be regarded as the background, the calculation is carried out by adopting a way of road network searching instead of grid traversal.
And step S4: and randomly selecting some station images, putting the station images into a trained model for power supply network rod identification, and checking an identification result.
Further, the specific steps of step S1 are:
step S11: according to the high-resolution satellite remote sensing image, drawing the proper position of the power supply network rod;
step S12: generating a minimum envelope rectangle for the position of a power supply network trunk by adopting a GIS tool;
step S13: considering the problem of file size, the original image needs to be cut according to the drawing area, and finally the shape file of the surface element, the KML file of the envelope rectangle and the TIFF file of the cut image are respectively stored.
Further, the specific steps of step S2 are:
step S21: and (3) putting the original image into a convolution network to extract characteristic information, wherein the loss function of FastRCNN is as follows:
L(p,u,t u ,v)＝L cls (p,u)+λ[u≥1]L loc (t u ,v) (1)
where p is the probability distribution p = (p) predicted by the classifier 0 ,…,p k ) U corresponds to the target truth class label, t u Regression parameters corresponding to class u predicted by bounding box regressor
v bounding box regression parameters for the true target (v) x ，v y ，v w ，v h )
Step S22: the feature image is used as a template and put into an RPN network to screen a boundary frame, then the boundary frame is further regressed to obtain a more accurate position, and coordinate information of the boundary frame is mapped onto a feature map, wherein the RPN network adopts the following loss function, and adopts individual discrimination as an agent task to construct a feature learning model in an automatic supervision mode, and parameters of the model are pre-trained:
wherein i refers to the index of an anchor in a mini-batch, p i Representing the estimated probability of whether or not anchor point i is an object. If the anchor point is a positive sample, true mark
Is 1 and the negative sample is 0.t is t i Vector of 4 parameterized coordinates representing a prediction bounding box, </or >>
Representing the real bounding box. Loss of classification L cls Is a logarithmic penalty covering both categories. We use { [ MEANS ] for regression loss>
Wherein R represents a robust smooth L1 loss. />
Representing the regression loss for the sample anchor point only, the others are ignored. The outputs of the cls and reg branches are respectively composed of { p i And t i And (9) composition.
And S23, putting the feature graph with the bounding box into an ROI Pooling layer to unify the dimensions of the bounding box and outputting a feature vector with a fixed size.
And S24, calculating the category in the fully-connected layer by using a Softmax classifier and performing regression of the bounding box again to obtain more accurate positioning.
Further, the specific steps of step S3 are:
step S31: performing initialization prediction by adopting the model obtained by S22 training according to the initially determined range of S11;
step S32: and when the station appears in the prediction result, establishing a root node. Furthermore, area grid searching is carried out near the station, sub-nodes are established when the railway exists nearby, all railway lines are further traversed through the area grid searching, the sub-nodes are returned when the power supply facility searching is finished, and prediction is carried out continuously along the advancing direction of the railway to finish the identification and extraction of the whole area power supply facility in sequence;
further, the specific steps of step S4 are:
step S41: and randomly selecting some station images as inspection data to be placed in the model for power supply network rod identification, and inspecting the confidence coefficient of the identification result.
Step S42: and (3) considering the influence of different landforms on the power supply facility identification monitoring, inputting the high-speed railway images under different landforms into the model, and checking the robustness of the model under the condition of coping with different influences.
Fig. 1 shows the overall architecture design of the present invention, which is divided into four parts S1, S2, S3, and S4 according to steps, and the four parts are sequentially performed in sequence. Each section has a separate output and is the input for the next section.
Fig. 2 shows an example of a minimum envelope rectangle containing the location of the mains pole, where the large rectangle is the envelope rectangle generated for the station and the small rectangle is the envelope rectangle generated for the mains pole.
Fig. 3 shows a grid diagram of a station and a power supply network pole outlined in the present invention, wherein the left side is a waiting hall of the station, the long strip is a track, and the middle is a platform. The original image is cut according to the drawing area, and the surface element is required to be converted into a raster file to be used as a cutting range.
In order to achieve efficient and stable target detection and meet the requirement of freeing human resources, the embodiment uses the relevant theory based on target detection in computer vision and builds a fast R-CNN deep machine learning model to solve the problems in the background art.
Compared with the traditional method based on manual interpretation and identification, the method based on computer vision for automatically identifying the power supply facilities through the target detection and deep learning model building has the advantages that the efficiency is greatly improved, the demand on manpower resources is reduced, time and labor are saved, and therefore, it is necessary to train the Faster R-CNN model and realize automatic identification and extraction of other areas by using 19-level resolution satellite images to obtain samples, and the method has great practical significance.
The present invention and its embodiments have been described above schematically, without limitation, and what is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. Therefore, if the person skilled in the art receives the teaching, without departing from the spirit of the invention, the person skilled in the art shall not inventively design the similar structural modes and embodiments to the technical solution, but shall fall within the scope of the invention.
Claims (5)
1. The intelligent identification method of the high-speed railway power supply facility based on the high-resolution remote sensing image is characterized by comprising the following steps: the method comprises the following steps:
step S1: selecting a high-resolution remote sensing image, and interpreting and drawing samples in the area around the high-speed train station;
step S2: setting initial parameters, calculating a loss function by using a sample with a boundary frame of the real position of the power supply network rod, and iterating and updating the parameter weight W and the offset value b in the process of participating in back propagation so as to finish the training of the model;
and step S3: calculating by adopting a way of road network searching;
and step S4: and randomly selecting some station images, putting the station images into a trained model for power supply network rod identification, and checking an identification result.
2. The intelligent identification method for the high-speed railway power supply facilities based on the high-resolution remote sensing image according to claim 1, characterized in that: the specific steps of the step S1 are as follows:
step S11: according to the high-resolution satellite remote sensing image, the position of the power supply network rod is sketched;
step S12: generating a minimum envelope rectangle for the position of a power supply network trunk by adopting a GIS tool;
step S13: and cutting the original image according to the drawing area, and finally respectively storing the shape file of the surface element, the KML file of the envelope rectangle and the TIFF file of the cut image.
3. The high-resolution remote sensing image-based intelligent identification method for the power supply facilities of the high-speed railway according to claim 2, wherein the method comprises the following steps: the specific steps of the step S2 are as follows:
step S21: placing the original image into a convolutional network FastRCNN to extract characteristic information, wherein the loss function of the FastRCNN is as follows:
L(p,u,t u ,v)＝L cls (p,u)+λ[u≥1]L loc (t u ,v) (1)
where p is the probability distribution predicted by the classifier, u corresponds to the target true class label, t u Corresponding to the regression parameters of the class u predicted by the boundary box regressor, and v corresponding to the boundary box regression parameters of the real target;
step S22: the feature image is used as a template and put into an RPN network to screen a boundary frame, then the boundary frame is further regressed to obtain a more accurate position, and coordinate information of the boundary frame is mapped onto a feature map, wherein the RPN network adopts the following loss function, and uses individual discrimination as an agent task to construct a feature learning model in an automatic supervision mode, and pre-trains parameters of the model:
wherein i refers to the index of an anchor in a mini-batch, p i An estimated probability representing whether the anchor point i is an object; if the anchor point is a positive sample, true mark
1, negative example 0; t is t i Vector of 4 parameterized coordinates representing a prediction bounding box, </or >>
Representing a real bounding box; loss of classification L cls Is a logarithmic loss covering both categories; for regression loss, use
Wherein R represents a robust smooth L1 loss; />
Represents the regression loss for sample anchor points only; the output of cls and reg branches are represented by { p } i And t i Is formed by the following steps;
s23, putting the feature map with the bounding box into an ROI Pooling layer to unify the dimensions of the bounding box and outputting a feature vector with a fixed size;
and S24, calculating the category in the fully-connected layer by using a Softmax classifier and performing regression of the bounding box again to obtain more accurate positioning.
4. The high-resolution remote sensing image-based intelligent identification method for the power supply facilities of the high-speed railway according to claim 3, wherein the method comprises the following steps: the specific steps of the step S3 are as follows:
step S31: performing initialization prediction by adopting the model obtained by S22 training according to the initially determined range of S11;
step S32: when a station appears in a prediction result, establishing a root node; and further, area grid searching is carried out near the station, sub-nodes are established when the railway exists nearby, all railway lines are further traversed through the area grid searching, the sub-nodes are returned when the power supply facility searching is completed, and the whole area power supply facility identification and extraction are sequentially completed by continuing to predict along the advancing direction of the railway.
5. The high-resolution remote sensing image-based intelligent identification method for the power supply facilities of the high-speed railway according to claim 4, wherein the method comprises the following steps: the specific steps of the step S4 are as follows:
step S41: randomly selecting some station images as test data to be placed in the model for power supply network rod identification, and testing the confidence coefficient of the identification result;
step S42: and (3) considering the influence of different landforms on the power supply facility identification monitoring, inputting the high-speed railway images under different landforms into the model, and checking the robustness of the model under the condition of coping with different influences.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211339322.XA CN115908897B (en) | 2022-10-28 | 2022-10-28 | Intelligent identification method for high-speed railway power supply facilities based on high-resolution remote sensing images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211339322.XA CN115908897B (en) | 2022-10-28 | 2022-10-28 | Intelligent identification method for high-speed railway power supply facilities based on high-resolution remote sensing images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115908897A true CN115908897A (en) | 2023-04-04 |
| CN115908897B CN115908897B (en) | 2023-09-12 |
Family
ID=86485289
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211339322.XA Active CN115908897B (en) | 2022-10-28 | 2022-10-28 | Intelligent identification method for high-speed railway power supply facilities based on high-resolution remote sensing images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115908897B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414795A (en) * | 2019-07-02 | 2019-11-05 | 华侨大学 | Method is influenced based on the newly-increased high-speed rail hinge accessibility for improving two moved further search methods |
| CN111091105A (en) * | 2019-12-23 | 2020-05-01 | 郑州轻工业大学 | Remote sensing image target detection method based on new frame regression loss function |
| CN111241994A (en) * | 2020-01-09 | 2020-06-05 | 中国交通通信信息中心 | Method for extracting remote sensing image rural highway desertification road section for deep learning |
| CN112287896A (en) * | 2020-11-26 | 2021-01-29 | 山东捷讯通信技术有限公司 | Unmanned aerial vehicle aerial image target detection method and system based on deep learning |
| CN112395975A (en) * | 2020-11-17 | 2021-02-23 | 南京泓图人工智能技术研究院有限公司 | Remote sensing image target detection method based on rotating area generation network |
| CN114694038A (en) * | 2022-04-11 | 2022-07-01 | 中国矿业大学 | High-resolution remote sensing image classification method and system based on deep learning |
-
2022
- 2022-10-28 CN CN202211339322.XA patent/CN115908897B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414795A (en) * | 2019-07-02 | 2019-11-05 | 华侨大学 | Method is influenced based on the newly-increased high-speed rail hinge accessibility for improving two moved further search methods |
| CN111091105A (en) * | 2019-12-23 | 2020-05-01 | 郑州轻工业大学 | Remote sensing image target detection method based on new frame regression loss function |
| CN111241994A (en) * | 2020-01-09 | 2020-06-05 | 中国交通通信信息中心 | Method for extracting remote sensing image rural highway desertification road section for deep learning |
| CN112395975A (en) * | 2020-11-17 | 2021-02-23 | 南京泓图人工智能技术研究院有限公司 | Remote sensing image target detection method based on rotating area generation network |
| CN112287896A (en) * | 2020-11-26 | 2021-01-29 | 山东捷讯通信技术有限公司 | Unmanned aerial vehicle aerial image target detection method and system based on deep learning |
| CN114694038A (en) * | 2022-04-11 | 2022-07-01 | 中国矿业大学 | High-resolution remote sensing image classification method and system based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115908897B (en) | 2023-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109493320B (en) | Remote sensing image road extraction method and system based on deep learning, storage medium and electronic equipment | |
| CN109376606B (en) | Power inspection image tower foundation fault detection method | |
| Zou et al. | CNN-based statistics and location estimation of missing components in routine inspection of historic buildings | |
| CN106919978A (en) | A kind of high ferro contact net support meanss parts recognition detection method | |
| CN106127204A (en) | A kind of multi-direction meter reading Region detection algorithms of full convolutional neural networks | |
| CN113592822B (en) | Insulator defect positioning method for electric power inspection image | |
| CN113177560A (en) | Universal lightweight deep learning vehicle detection method | |
| CN112990086A (en) | Remote sensing image building detection method and device and computer readable storage medium | |
| CN110910440B (en) | Power transmission line length determination method and system based on power image data | |
| CN115409069A (en) | Village and town building identification method, classification method, device, electronic equipment and medium | |
| CN116863274A (en) | Semi-supervised learning-based steel plate surface defect detection method and system | |
| CN113297915A (en) | Insulator recognition target detection method based on unmanned aerial vehicle inspection | |
| CN112001293A (en) | Remote sensing image ground object classification method combining multi-scale information and coding and decoding network | |
| CN115830535A (en) | Method, system, equipment and medium for detecting accumulated water in peripheral area of transformer substation | |
| Hassan et al. | Real-time road cracks detection based on improved deep convolutional neural network | |
| CN113902792A (en) | Building height detection method and system based on improved RetinaNet network and electronic equipment | |
| CN114037836A (en) | Method for applying artificial intelligence recognition technology to three-dimensional power transmission and transformation engineering measurement and calculation | |
| CN115908897B (en) | Intelligent identification method for high-speed railway power supply facilities based on high-resolution remote sensing images | |
| CN109919990B (en) | Forest height prediction method by using depth perception network and parallax remote sensing image | |
| CN111428191A (en) | Antenna downward inclination angle calculation method and device based on knowledge distillation and storage medium | |
| CN115937492A (en) | Transformer equipment infrared image identification method based on feature identification | |
| Zhang et al. | Pavement crack detection based on deep learning | |
| CN115731560A (en) | Slot line identification method and device based on deep learning, storage medium and terminal | |
| CN112597948B (en) | Urban land utilization change prediction method | |
| Wang et al. | The building area recognition in image based on faster-RCNN |
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 |