CN114036822A - Rapid thermal model construction method based on neural network - Google Patents
Rapid thermal model construction method based on neural network Download PDFInfo
- Publication number
- CN114036822A CN114036822A CN202111238644.0A CN202111238644A CN114036822A CN 114036822 A CN114036822 A CN 114036822A CN 202111238644 A CN202111238644 A CN 202111238644A CN 114036822 A CN114036822 A CN 114036822A
- Authority
- CN
- China
- Prior art keywords
- thermal
- model
- neural network
- level
- rapid thermal
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/39—Circuit design at the physical level
- G06F30/398—Design verification or optimisation, e.g. using design rule check [DRC], layout versus schematics [LVS] or finite element methods [FEM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/14—Details relating to CAD techniques related to nanotechnology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a rapid thermal model construction method based on a neural network, which comprises the following steps: s1, establishing a hot link model; s2, acquiring and preprocessing thermal model modeling sample data; s3, establishing and training a neural network kernel model; s4, rapid thermal model packaging, etc. The method solves the common problems of large number of grids, difficult convergence and long consumed time in the thermal evaluation process of the complex system, and can greatly improve the thermal simulation efficiency and the like on the premise of ensuring the simulation progress.
Description
Technical Field
The invention relates to the technical field of electronic equipment thermal management, in particular to a rapid thermal model construction method based on a neural network.
Background
With the development of electronic products towards miniaturization and integration, electronic heat dissipation has become a bottleneck factor restricting the development of equipment. The current thermal simulation is mainly based on mature commercial software, the modeling speed is slow, and the grid division method is complex. Meanwhile, with the increase of the heat flux density, the number of the grids of the thermal simulation with the complex structure is large, and rapid convergence is difficult, so that the simulation time is long, and therefore, common simulation software is not suitable for certain application scenes needing rapid evaluation of the feasibility of the thermal design.
In the prior art, Chinese patent with publication number CN110083125A provides a machine tool thermal error modeling method based on deep learning, and the invention can effectively estimate the machine tool thermal error change trend; the Chinese patent application with publication number CN111126827A provides a method for constructing an input-output accounting model based on a BP artificial neural network, and the method acquires an input-output table of city scale by constructing an input-output BP artificial neural network model; chinese patent publication No. CN109739181A proposes a machine tool spindle thermal error modeling detection method based on a detection neural network. However, the simulation model constructed based on the neural network algorithm and other algorithms is mainly used for result prediction, error compensation and the like, and does not relate to the construction of the rapid thermal model based on the neural network algorithm.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a rapid thermal model construction method based on a neural network, solves the common problems of large number of grids, difficult convergence and long time consumption in the thermal evaluation process of a complex system, and can greatly improve the thermal simulation efficiency and the like on the premise of ensuring the simulation progress.
The purpose of the invention is realized by the following scheme:
a rapid thermal model construction method based on a neural network comprises the following steps:
s1, establishing a hot link model;
s2, acquiring and preprocessing thermal model modeling sample data;
s3, establishing and training a neural network kernel model;
and S4, packaging by rapid thermal model.
Further, in step S1, the method includes the sub-steps of: extracting key thermal parameters of a chip level, a module level and a system level from the sample data, and constructing thermal resistance and thermal capacity models of all levels so as to construct a one-dimensional thermal network model.
Further, in step S2, the preprocessing includes the sub-steps of:
s21, sequentially sorting the sensitivity of each influence factor of the acquired data;
and S22, increasing the number of samples of factors which have important influence on the thermal evaluation result after being sorted in the step S21, and further acquiring more comprehensive and effective modeling sample data.
Further, in step S3, the thermal model modeling sample data includes sample data of chip level, module level, system level thermal resistance, and thermal capacitance model.
Further, in step S4, the method includes the steps of:
and during packaging, a model interface file which is in butt joint with professional thermal simulation software and a tool is added, and a model dynamic link library file which can be directly called is generated by compiling and linking, can be correctly mapped to the neural network model and is called to drive to complete a calculation process.
Further, in step S3, the method includes the sub-steps of:
respectively carrying out batch loading on the sample data of the chip-level, module-level, system-level thermal resistance and thermal capacity models into a neural network with a determined structure for training, and respectively obtaining a neural network model representing the behavior characteristics of the chip-level, module-level, system-level thermal resistance and thermal capacity models; the input of each neural network model is a parameter influencing each level of thermal resistance and thermal capacity models.
Further, in step S3, a BP neural network is used for modeling, the number of hidden layers is usually 1-2, hidden layer neurons use Sigmoid type transfer functions, and output layer neurons use purelin type transfer functions.
Further, the parameters include the depth of the flow channel, the width of the flow channel and the flow rate, and the output of each neural network model is the corresponding thermal resistance and thermal capacity value.
Further, the rapid thermal model comprises a microchannel heat sink rapid thermal model.
The beneficial effects of the invention include:
the embodiment of the invention is suitable for effectively processing multivariable and nonlinear basic data and establishing the mapping relation among the data, and utilizes the characteristics of multivariable, nonlinearity and the like of the thermal simulation model, so that the application characteristics of the thermal simulation model are matched with the application characteristics of a neural network algorithm, and the common problems of large number of grids, difficult convergence and long consumed time in the thermal evaluation process of a complex system are solved. Specifically, the method trains input parameters and output result data of an original model based on a neural network algorithm, and performs customized packaging on the model, so as to construct a hot link black box model. The processing method can accurately fit the complex nonlinear mapping relation of the model, does not need to know the internal structure and the working mechanism of the thermal model, can be directly called by professional thermal simulation software and tools, and can greatly improve the thermal simulation efficiency on the premise of ensuring the simulation progress.
The embodiment of the invention establishes a neural network mathematical mapping relation reflecting multi-dimensional thermal characteristics by analyzing the characteristics and influence parameters of the heat dissipation system on the basis of sample data formed by theoretical analysis calculation, thermal simulation, test verification and the like, thereby accurately characterizing and modeling the heat dissipation system. Compared with the traditional modeling method, the embodiment of the invention can quickly convert the test and simulation data into the quick thermal model compatible with mature professional thermal simulation software and tools without ascertaining the internal structure and the working principle of the system, thereby being capable of carrying out quick thermal simulation in the professional simulation software and tools. When a designer carries out thermal simulation on a complex heat dissipation system, the designer can quickly solve an accurate simulation result by only inputting simulation parameters. Taking the heat dissipation evaluation of the microchannel heat sink as an example, the dimension of the microchannel heat sink belongs to the micron level, the grid number of a common thermal simulation model reaches the million level, the simulation speed is very slow, and the rapid thermal model simulation based on the neural network only needs 30 seconds.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a rapid thermal model construction based on a neural network in an embodiment of the present invention;
FIG. 2 is a schematic diagram of a basic structure of a micro-channel heat sink according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a thermal model interface in an embodiment of the present invention;
FIG. 4 is a schematic diagram of a thermal resistance model in an embodiment of the invention;
FIG. 5 is a graph illustrating performance of model training results according to an embodiment of the present invention;
FIG. 6 is a graph of model prediction result performance indicators in an embodiment of the present invention;
fig. 7 is a schematic view of a rapid thermal model package interface according to an embodiment of the present invention.
Detailed Description
All features disclosed in all embodiments in this specification, or all methods or process steps implicitly disclosed, may be combined and/or expanded, or substituted, in any way, except for mutually exclusive features and/or steps.
Examples
The invention is applied to the construction of the rapid thermal model of the micro-channel radiator, the micro-channel radiator is in a micron scale, and the system shunt network is in a hundred-millimeter scale, so the spatial scale can span up to 6 orders of magnitude, the simulation time consumption is long and the simulation difficulty is high by adopting conventional commercial software, and the invention is very suitable for the rapid thermal model based on the neural network to carry out high-efficiency evaluation. The rapid thermal model construction process based on the neural network is shown in figure 1.
The simulation system of the micro-channel radiator mainly comprises a four-way shunt network, a module-level assembly, the micro-channel radiator and the like, and the basic composition architecture of the simulation system is shown in fig. 2.
The method comprises the following steps: establishing a thermal link model
Taking the steady-state thermal simulation of the microchannel heat sink as an example for analysis, the input parameters include physical parameters (such as the depth and width of the flow channel of the microchannel heat sink) and environmental parameters (such as the inlet flow and the ambient temperature) of the model, and the thermal model interface is shown in fig. 2. The whole thermal network model comprises three-stage thermal resistances, namely chip thermal resistance, micro-channel radiator thermal resistance and system shunt network thermal resistance. The thermal resistance model is shown in fig. 3 and 4.
(1) Chip vertical average thermal resistance Rchip
In modeling, thermal resistance calculated by corresponding average temperatures of the upper surface and the lower surface of the chip is defined, namely:
wherein R iscIs the average thermal resistance of the chip, and QcFor power consumption, Δ T is the temperature difference between the upper and lower surfaces of the chip.
The thermal resistance is only relevant to the chip and is therefore a determined value for a determined chip. The function R related to the chip thermal resistance can be obtained by carrying out neural network mathematical modeling aiming at simulation and test sample datac(x1,x2...q,TInto) Thus, it is possible to obtain:
Tj=Ti+P.Rc
wherein x is1、x2、x3Is the physical parameters of the micro-channel radiator such as the depth of a flow channel, the width of the flow channel and the like, q is the inlet flow, and T isIntoIs the inlet temperature.
(2) Distributed thermal resistance R on chip surfacexy
RxyDefined as the value of any position coordinate (x, y) on the chip surface relative to the average thermal resistance, the physical parameters of the micro-channel radiator, the distribution of the surface thermal resistance, the flow rate of the micro-channel and the temperature of the inlet liquidDegree has an effect on the thermal resistance, so Rchip(x,y,q,T)=fANN(x,y,q,T)+Rchip_mean。
(3) Module level thermal resistance model Rmc
Because complete parameterization of a module level is difficult to realize, typical parameters are selected as simulation parameters (designPar) of a module level thermal resistance model, and a mapping relation between the parameters and module level equivalent thermal resistance, namely Rmc(designPar)ΔPremc(designPar)]=fANN(designPar)。
(4) System level shunt network modeling
The system shunting of this embodiment is to shunt four ways through the flow distribution plate, and in order to evaluate the shunting flow and temperature of each module under different volume flows and ambient temperatures, a system level shunting network needs to be modeled. Based on the simulation data, the following functional relationship is established.
Wherein q isxFor shunting flow to a certain path, TxThe temperature of the heat sink entering a certain micro-channel.
Step two: sample data acquisition
The thermal simulation model mainly comprises a four-way shunt network, a module component, a micro-channel radiator and the like, and main factors influencing thermal analysis mainly comprise liquid inlet temperature, inlet volume flow, flow channel width, flow channel depth, liquid inlet and outlet diameter, tooth space width ratio, total flow channel width and the like.
The acquisition of sample data is mainly based on mature thermal simulation software. After the first batch of sample data is obtained, analyzing the influence of each input factor on the highest temperature and flow resistance (inlet-outlet pressure difference) of the chip based on statistical software, carrying out sensitivity analysis on simulation data, and sequencing the sensitivity of each influence factor. Because the magnitude difference of the data is large, the data needs to be normalized, and then regression analysis is performed on the processed data. According to the analysis result, the ranking of the influence of each factor on the highest temperature of the chip is as follows: the liquid inlet temperature is larger than the inlet volume flow, the flow channel width is larger than the flow channel depth, the liquid inlet and outlet diameter is larger than the tooth space width ratio and the total width of the flow channel. Based on the method, the simulation sample size of the factors such as the liquid inlet temperature, the inlet volume flow and the like is increased, and more comprehensive and effective modeling sample data can be obtained
Step three: neural network establishment and training
The embodiment of the invention mainly adopts the BP neural network for modeling. The function of the neural network training toolbox of mature commercial software can meet the application requirement of the embodiment, so the neural network training toolbox can be used for training. In order to make the training process more compatible with the contents of the present invention, a certain customized development of the toolbox is required. After neural network training, the performance indexes of the model training result and the performance indexes of the model prediction result are shown in fig. 5 and 6.
Step four: rapid thermal model encapsulation
A complete and reliable neural network model has been built based on the above-described process, and then the neural network model is encapsulated based on a programming language. The encapsulated model interface is schematically shown in fig. 7. The packaged software interface mainly comprises a main menu bar, a shortcut toolbar, a main project view, a parameter setting window, a main interface, a message window and the like. The main menu column is an inlet of various functions, the parameter input column is a detailed setting of simulation parameters, the main interface mainly delivers and displays simulation results and can also inquire simulation objects, and in the actual application process, a designer can output a required simulation result by one key only by setting the simulation parameters in the parameter setting column.
The invention provides a method for constructing a rapid thermal model based on a neural network, which is based on sample data formed by theoretical analysis calculation, thermal simulation, test verification and the like, and establishes a neural network mathematical mapping relation reflecting multi-dimensional thermal characteristics by analyzing the characteristics and influence parameters of a heat dissipation system, thereby accurately characterizing and modeling the heat dissipation system. Compared with the traditional modeling method, the method can quickly convert the test and simulation data into the quick thermal model compatible with mature professional thermal simulation software and tools without ascertaining the internal structure and the working principle of the system, and further can carry out quick thermal simulation in the professional simulation software and tools. When a designer carries out thermal simulation on a complex heat dissipation system, the designer can quickly solve an accurate simulation result by only inputting simulation parameters. Taking the heat dissipation evaluation of the microchannel heat sink as an example, the dimension of the microchannel heat sink belongs to the micron level, the grid number of a common thermal simulation model reaches the million level, the simulation speed is very slow, and the rapid thermal model simulation based on the neural network only needs 30 seconds.
The parts not involved in the present invention are the same as or can be implemented using the prior art.
The above-described embodiment is only one embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be easily made based on the application and principle of the present invention disclosed in the present application, and the present invention is not limited to the method described in the above-described embodiment of the present invention, so that the above-described embodiment is only preferred, and not restrictive.
Other embodiments than the above examples may be devised by those skilled in the art based on the foregoing disclosure, or by adapting and using knowledge or techniques of the relevant art, and features of various embodiments may be interchanged or substituted and such modifications and variations that may be made by those skilled in the art without departing from the spirit and scope of the present invention are intended to be within the scope of the following claims.
Claims (9)
1. A rapid thermal model construction method based on a neural network is characterized by comprising the following steps:
s1, establishing a hot link model;
s2, acquiring and preprocessing thermal model modeling sample data;
s3, establishing and training a neural network kernel model;
and S4, packaging by rapid thermal model.
2. The neural network-based rapid thermal model building method according to claim 1, comprising, in step S1, the sub-steps of: extracting key thermal parameters of a chip level, a module level and a system level from the sample data, and constructing thermal resistance and thermal capacity models of all levels so as to construct a one-dimensional thermal network model.
3. The neural network-based rapid thermal model building method according to claim 1, wherein in step S2, the preprocessing includes the sub-steps of:
s21, sequentially sorting the sensitivity of each influence factor of the acquired data;
and S22, increasing the number of samples of factors which have important influence on the thermal evaluation result after being sorted in the step S21, and further acquiring more comprehensive and effective modeling sample data.
4. The neural network-based rapid thermal model building method according to claim 1, wherein in step S3, the thermal model modeling sample data includes sample data of chip-level, module-level, system-level thermal resistance, thermal capacity model.
5. The neural network-based rapid thermal model building method according to claim 1, comprising, in step S4, the steps of: and during packaging, a model interface file which is in butt joint with professional thermal simulation software and a tool is added, and a model dynamic link library file which can be directly called is generated by compiling and linking, can be correctly mapped to the neural network model and is called to drive to complete a calculation process.
6. The neural network-based rapid thermal model building method according to claim 4, comprising, in step S3, the sub-steps of: respectively carrying out batch loading on the sample data of the chip-level, module-level, system-level thermal resistance and thermal capacity models into a neural network with a determined structure for training, and respectively obtaining a neural network model representing the behavior characteristics of the chip-level, module-level, system-level thermal resistance and thermal capacity models; the input of each neural network model is a parameter influencing each level of thermal resistance and thermal capacity models.
7. The method for rapid thermal model construction based on neural network as claimed in claim 4, wherein in step S3, modeling is performed by using BP neural network, the number of hidden layers is usually 1-2, hidden layer neurons use Sigmoid type transfer function, and output layer neurons use purelin type transfer function.
8. The method of claim 6, wherein the parameters include channel depth, channel width, and flow rate, and the output of each neural network model is corresponding thermal resistance and thermal capacitance values.
9. The neural network-based rapid thermal model construction method according to any one of claims 1-8, wherein the rapid thermal model comprises a microchannel heat sink rapid thermal model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111238644.0A CN114036822B (en) | 2021-10-25 | 2021-10-25 | Quick thermal model construction method based on neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111238644.0A CN114036822B (en) | 2021-10-25 | 2021-10-25 | Quick thermal model construction method based on neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114036822A true CN114036822A (en) | 2022-02-11 |
CN114036822B CN114036822B (en) | 2023-07-04 |
Family
ID=80141803
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111238644.0A Active CN114036822B (en) | 2021-10-25 | 2021-10-25 | Quick thermal model construction method based on neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114036822B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101201907A (en) * | 2007-10-24 | 2008-06-18 | 中山大学 | Hot comfortableness simulation model based on fuzzy neural network |
CN112231985A (en) * | 2020-11-04 | 2021-01-15 | 中国电子科技集团公司第二十九研究所 | Radio frequency filter modeling method |
CN112926258A (en) * | 2021-02-05 | 2021-06-08 | 西安电子科技大学 | Method for predicting junction temperature of semiconductor device based on BP neural network model |
CN113224822A (en) * | 2021-05-26 | 2021-08-06 | 北京小米移动软件有限公司 | Charging control method and device and storage medium |
WO2021174907A1 (en) * | 2020-03-03 | 2021-09-10 | 华中科技大学 | Neural network-based igbt junction temperature prediction method |
-
2021
- 2021-10-25 CN CN202111238644.0A patent/CN114036822B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101201907A (en) * | 2007-10-24 | 2008-06-18 | 中山大学 | Hot comfortableness simulation model based on fuzzy neural network |
WO2021174907A1 (en) * | 2020-03-03 | 2021-09-10 | 华中科技大学 | Neural network-based igbt junction temperature prediction method |
CN112231985A (en) * | 2020-11-04 | 2021-01-15 | 中国电子科技集团公司第二十九研究所 | Radio frequency filter modeling method |
CN112926258A (en) * | 2021-02-05 | 2021-06-08 | 西安电子科技大学 | Method for predicting junction temperature of semiconductor device based on BP neural network model |
CN113224822A (en) * | 2021-05-26 | 2021-08-06 | 北京小米移动软件有限公司 | Charging control method and device and storage medium |
Non-Patent Citations (3)
Title |
---|
YAN-LIANG DONG等: "Influence of the thermal effect on the sealing performance of the hydraulic combined dynamic seal", vol. 48, no. 18, pages 9 - 15 * |
徐榕青;张晏铭;王辉;李杨;庞婷;: "面向微波组件工艺失效分析的大数据建模技术", no. 09, pages 244 - 248 * |
林佳;刘云峰;尹本浩;谭继勇;陈显才;: "一种复合相变热沉设计与散热性能分析", no. 04, pages 80 - 86 * |
Also Published As
Publication number | Publication date |
---|---|
CN114036822B (en) | 2023-07-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102214261B (en) | Method for automatic check of process data of printed circuit board | |
CN109783970B (en) | Thermal analysis method for reliability simulation analysis of electronic product | |
CN110222828B (en) | Unsteady flow field prediction method based on hybrid deep neural network | |
CN107844424A (en) | Model-based testing system and method | |
CN107545349A (en) | A kind of Data Quality Analysis evaluation model towards electric power big data | |
CN107346286A (en) | A kind of Software Defects Predict Methods based on core principle component analysis and extreme learning machine | |
US20230195986A1 (en) | Method for predicting delay at multiple corners for digital integrated circuit | |
Di Filippo et al. | Generative Design for project optimization (S). | |
TWI689724B (en) | Chip temperature computation method and chip temperature computation device | |
CA2652710A1 (en) | Pruning-based variation-aware design | |
CN112257303B (en) | Temperature stabilization time testing method based on thermal simulation model | |
US5675521A (en) | Multichip module analyzer | |
CN113177290A (en) | Satellite component temperature field prediction method based on depth proxy model normalization | |
CN117077526A (en) | Digital twinning-based radar TR module temperature prediction method | |
CN112580288B (en) | Semiconductor device characteristic modeling method and system based on multi-gradient neural network | |
CN114036822B (en) | Quick thermal model construction method based on neural network | |
Luo et al. | A fixture design retrieving method based on constrained maximum common subgraph | |
CN108399284A (en) | It is a kind of about subtracted based on deviation big data Trading Model analysis and restorative procedure | |
Le et al. | A generative network model for product evolution | |
CN103488842B (en) | A kind of reliability automatic method for predicting of typical discrete device based on Saber | |
CN114021514B (en) | Method for simulating and screening bottleneck units through SPICE voltage or temperature scanning | |
CN115455891A (en) | Method and system for predicting parasitic parameters in simulation process before integrated circuit simulation | |
CN114266178A (en) | Power device health state assessment method and system | |
Williams et al. | Deriving metamodels to relate machine learning quality to design repository characteristics in the context of additive manufacturing | |
CN112950279A (en) | Accurate marketing strategy model construction method and device based on machine learning |
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 |