CN110363208B - Multi-parameter fusion stacker track fault detection and early warning method - Google Patents

Multi-parameter fusion stacker track fault detection and early warning method Download PDF

Info

Publication number
CN110363208B
CN110363208B CN201810312659.9A CN201810312659A CN110363208B CN 110363208 B CN110363208 B CN 110363208B CN 201810312659 A CN201810312659 A CN 201810312659A CN 110363208 B CN110363208 B CN 110363208B
Authority
CN
China
Prior art keywords
data
fusion
interval
feature
early warning
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810312659.9A
Other languages
Chinese (zh)
Other versions
CN110363208A (en
Inventor
王健
赵良德
陈晨
谢乐天
刘洪亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Nanrui Jiehong Technology Co ltd
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Anhui Electric Power Co Ltd
NARI Group Corp
Original Assignee
Beijing Nanrui Jiehong Technology Co ltd
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Anhui Electric Power Co Ltd
NARI Group Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Nanrui Jiehong Technology Co ltd, State Grid Corp of China SGCC, Electric Power Research Institute of State Grid Anhui Electric Power Co Ltd, NARI Group Corp filed Critical Beijing Nanrui Jiehong Technology Co ltd
Priority to CN201810312659.9A priority Critical patent/CN110363208B/en
Publication of CN110363208A publication Critical patent/CN110363208A/en
Application granted granted Critical
Publication of CN110363208B publication Critical patent/CN110363208B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • 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
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/20—Administration of product repair or maintenance

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Business, Economics & Management (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Genetics & Genomics (AREA)
  • Physiology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Measurement Of Radiation (AREA)
  • Machines For Laying And Maintaining Railways (AREA)

Abstract

The invention discloses a multi-parameter fusion stacker track fault detection and early warning method. The method comprises the following steps: data preprocessing, feature extraction, feature fusion, fault diagnosis and early warning. The data preprocessing comprises the following steps: smoothing the original data S by using interval sampling to obtain a result SI'; determining the initial position and the sampling interval of the stacker are respectively as follows: start and interval; thereby forming an initial form of interval sampling and being beneficial to the smoothness of a subsequent interval; the method has the advantages that the interference of accidental faults and invisible faults is solved by an important step of data processing during interval sampling, the smoothness of the track data is realized, and the method lays a cushion for feature extraction in the later period. The invention not only reduces the parameter dimension by the feature extraction of the LDA algorithm and reduces the operation cost, but also improves the accuracy of fault diagnosis by the parameter fusion by optimizing the parameter weighting of the genetic algorithm; the track fault detection device can reliably detect track faults and warn workers to maintain in time, mechanical damage of a factory is reduced, and cost is saved.

Description

Multi-parameter fusion stacker track fault detection and early warning method
Technical Field
The invention relates to a stacker track fault detection and early warning method, in particular to a multi-parameter fusion stacker track fault detection and early warning method.
Background
In recent years, with the development of the logistics industry and the computer information technology, the scale of modern enterprises is continuously enlarged and competition is increasingly intensified, the market puts new demands on the logistics system of the enterprises, and the automatic stereoscopic warehouse is more and more concerned and widely applied. The stacker is the most important carrying, lifting and stacking equipment of the automatic stereoscopic warehouse, and is one of key equipment for judging whether the stereoscopic warehouse can meet design requirements and embodying the advantages of the stereoscopic warehouse.
The stacker is the most critical part of the automatic warehouse and is a link and a bridge for the operation of the whole system, and whether the stacker operates well or not has the vital influence on the normal operation of the whole automatic warehouse. However, the rail is used as a decisive factor for ensuring the normal and stable operation of the stacker, the failure of the rail can cause the unstable operation of the stacker, the precision of goods storage and taking is reduced, and the possibility of serious mechanical failure is increased, so the failure detection and early warning can timely discover and maintain the failure of the rail, thereby ensuring the safe operation of the whole stacker system.
China still finds the fault detection of the stacker in a primary stage, and many technologies aim at other parts of the stacker and rarely detect the faults of the rails. At present, some track detection algorithms mainly rely on single parameters, and have the problems of high calculation complexity, low diagnosis precision, easy misjudgment and the like. In view of this, the method can solve the problem of low efficiency through track multi-parameter fusion, and has the advantages of small data volume and simple calculation after LDA feature extraction.
In conclusion, the invention designs a multi-parameter fusion stacker track fault detection and early warning method.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a multi-parameter fusion stacker track fault detection and early warning method, which not only performs feature extraction and parameter dimension reduction through an LDA algorithm and reduces the operation cost, but also performs parameter fusion through optimizing parameter weighting through a genetic algorithm and improves the fault diagnosis accuracy. The method can reliably detect the track fault and warn workers to maintain in time, reduces the mechanical damage of a factory, saves the cost, and has certain economic benefit and engineering practicability.
In order to achieve the purpose, the invention is realized by the following technical scheme: a multi-parameter fusion stacker track fault detection and early warning method comprises the following processes: data preprocessing, feature extraction, feature fusion, fault diagnosis and early warning.
Preferably, the data preprocessing comprises the following steps: in the actual operation process, the operation interval is repeated and contains the starting current and the stopping current, so that the original data S is smoothed by using interval sampling to obtain a result SI'. Determining the initial position and the sampling interval of the stacker are respectively as follows: start and interval. Therefore, the initial form of interval sampling is formed, and the subsequent interval smoothing is facilitated. The method has the advantages that the interference of accidental faults and invisible faults is solved by an important step of data processing during interval sampling, the smoothness of the track data is realized, and the method lays a cushion for feature extraction in the later period.
Preferably, the feature extraction is based on LDA multi-parameter feature extraction, and the LDA is used for performing feature extraction on the smoothed data SI' so as to achieve the purpose of dimension reduction. Firstly, determining an original central point of data as m, determining a distance between a data point and a classification center as D, and setting a data classification number as n. The LDA is utilized to extract fault characteristics of a plurality of acquired track parameters, and the characteristics are obtained as lambda (lambda ═ lambda [ () 1 ,λ 2 ,…,λ N ) And N represents the number of parameters, so as to prepare for later-stage feature fusion.
Preferably, the feature fusion is a multi-parameter weighted fusion in which a weight coefficient is optimized by a Genetic Algorithm (GA), and the feature λ fusion weight coefficient μ ═ μ by the Genetic Algorithm (GA) 1 ,μ 2 ,…,μ N ) Optimizing to obtain the optimal weight coefficient mu OPT ＝(μ OPT1 ,μ OPT2 ,…μ OPTN ) To achieve the optimal fault feature fusion
Figure BDA0001622884850000021
And the obtained fusion characteristic fault diagnosis precision is maximized and serves as a premise of later-stage fault diagnosis. Setting the genetic number as M, the genetic algebra as L, the cross probability as px, the mutation probability as pm, the variable binary number as PR, the surrogate furrow as GGA and the objective function as f.
Preferably, the fault diagnosis and early warning refers to fault diagnosis and early warning based on a BP neural network, the neural network is an intelligent method, the method is used for fault classification, manual experience influence can be eliminated, and diagnosis precision is improved. The fault signature obtained previously is used as the BP neural network input, and the fault type is used as the output. Therefore, the number of nodes of the input layer is set to be I, the number of nodes of the output layer is set to be O, and the number of nodes of the hidden layer is set to be H. Firstly, training BP by using a training set, and learning to obtain data characteristics. Secondly, testing is carried out by utilizing the test set, and if the diagnosed fault reaches the specified level, alarming is carried out.
According to the data preprocessing of the invention, the original data has a repeated redundancy phenomenon, so that the data needs to be preprocessed. And performing feature extraction on the preprocessed data by utilizing the LDA, thereby performing dimension reduction processing on the data. And seeking an optimal characteristic weight coefficient by using a genetic algorithm, and carrying out weighting processing on a plurality of characteristics to complete multi-parameter characteristic fusion. And determining the number of BP neural network nodes, the sizes of a training set and a test set, diagnosing faults and early warning.
The invention has the following beneficial effects:
1. the calculation method has high efficiency
After the LDA is used for carrying out dimensionality reduction feature extraction on the data, the data dimensionality is reduced, and the calculated amount is greatly reduced. Secondly, intelligent classification is carried out by using a BP neural network with the simplest structure. Most of the traditional fault detection methods are analyzed from the stacker structure and the control method so as to diagnose the fault, and the method analyzes and excavates data, so that the calculation is simple and the efficiency is high.
2. The diagnosis precision is high
Compared with other methods, the method has higher diagnosis precision. Firstly, interval smoothing is carried out, and interference of accidental faults is eliminated; secondly, orbit diagnostics no longer rely on only a single parameter. The probability of carrying out error judgment of the rail fault diagnosis by using a single parameter is far greater than that of multi-parameter fusion diagnosis, so that all parameters are comprehensively considered for carrying out comprehensive diagnosis, the error judgment probability is reduced, and the diagnosis precision is improved.
3. Strong modularization and relative independence
The modules of data preprocessing, feature extraction, feature fusion and intelligent diagnosis are mutually independent, so that the method can be modularized, portable and changeable.
Drawings
The invention is described in detail below with reference to the drawings and the detailed description;
FIG. 1 is an overall flow chart of the present invention;
FIG. 2 is a flow chart of data smoothing according to the present invention;
FIG. 3 is a LDA feature extraction flow chart of the present invention;
FIG. 4 is a feature fusion flow diagram of the present invention;
FIG. 5 is a flowchart of the BP neural network fault diagnosis of the present invention;
FIG. 6 is a diagram of the BP neural network structure according to the present invention.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.
Referring to fig. 1 to 6, the following technical solutions are adopted in the present embodiment: a multi-parameter fusion stacker track fault detection and early warning method comprises the following processes: data preprocessing, feature extraction, feature fusion, fault diagnosis and early warning. The specific method comprises the following steps:
the first step of data interval smoothing is shown in fig. 2. The range of the stacker track is 0-24 rows, and the length is 0m-24 m. First, the initial position start is set to 0mm and the interval sampling interval is set to 100mm, and therefore the interval size a is set to [ start, start + interval ═ a ═ start, start + interval ═ c]. Sorting the original data according to ascending order of distance, and countingThe number of current data to the section i is index and is represented as SI i ＝(s 1 ,s 2 …,s index ) Therefore, the data can be obtained SI after the interval i is smoothed i ', as follows:
Figure BDA0001622884850000041
therefore, the smoothed original data S yields a data set of: SI '═ SI' 1 ,…,SI' O )
The second step is feature extraction using LDA, as shown in fig. 3. In this method, the mean of the t-th class samples is first calculated as follows:
Figure BDA0001622884850000042
wherein r is i Indicates the number of samples belonging to the category t, and x indicates the sample.
Second, the overall sample mean is calculated as follows:
Figure BDA0001622884850000043
wherein n represents the number of categories, x i The ith sample is represented.
Defining the inter-class dispersion matrix and the intra-class dispersion matrix as follows:
Figure BDA0001622884850000051
Figure BDA0001622884850000052
here, a Fisher discrimination criteria expression is introduced:
Figure BDA0001622884850000053
the third step is the selection of the optimal feature fusion weight coefficients, as shown in fig. 3. In the method, a genetic algorithm is adopted to optimize the weighting weight coefficient. Setting genetic algebra L as 100, px as 0.97, variation probability pm as 0.001, number of individuals as 50, channel GGA as 0.90, weight coefficient mu as (mu) 1 ,μ 2 ,…,μ N ) The feature fusion calculation is as follows:
Figure BDA0001622884850000054
and set the objective function as follows:
f＝max(PRE)
where PRE is referred to as classification accuracy.
Optimizing to obtain the optimal weight coefficient mu OPT ＝(μ OPT1 ,μ OPT2 ,…μ OPTN )。
And fourthly, performing feature recognition and fault diagnosis by using the BP neural network, as shown in figure 5. The BP neural network structure is as shown in fig. 6, the number of nodes in the input layer is set to be I, which is the dimension of the fault feature, and represents five fault features, and the number of nodes in the output layer O is set to be 3 to represent 5 faults, so the number of hidden nodes is as follows:
Figure BDA0001622884850000055
the network function is defined as follows:
Figure BDA0001622884850000056
wherein Q ═ Q (Q) 1 ,…,q V ) Representing the weight coefficient, b representing the bias.
In this method, an error function is defined as:
Figure BDA0001622884850000057
wherein E is trainingNumber of samples, d (i) corresponding to input λ OPTi The desired output of (c).
The weight and bias updating formula of each iteration is as follows:
Figure BDA0001622884850000061
Figure BDA0001622884850000062
where α is the learning rate, its range (0, 1).
And obtaining the optimal network weight coefficient through the training, and classifying the test data to diagnose the fault and alarm.
The fault detection and early warning method of the embodiment is mainly used for detecting the track fault of the stacker and carrying out early warning so as to ensure timely maintenance. The method not only reduces the parameter dimension by feature extraction through the LDA algorithm and reduces the operation cost, but also improves the accuracy of fault diagnosis by optimizing parameter weighting through the genetic algorithm and performing parameter fusion. The method can reliably detect the track fault and warn workers to maintain in time, reduces the mechanical damage of a factory, saves the cost, and has certain economic benefit and engineering practicability.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (1)

1. A multi-parameter fusion stacker track fault detection and early warning method is characterized by comprising the following processes: data preprocessing, feature extraction, feature fusion, fault diagnosis and early warning;
the data preprocessing comprises the following steps: smoothing the original data S by using interval sampling to obtain a result SI';
Figure FDA0003722332980000011
counting the number of current data falling to the section i as index;
determining the initial position and the sampling interval of the stacker are respectively as follows: start and interval; thereby forming an initial form of interval sampling and being beneficial to the smoothness of a subsequent interval; the method has the advantages that the interference of accidental faults and invisible faults is solved by an important step of data processing during interval sampling, the smoothness of the track data is realized, and the method lays a cushion for later-stage feature extraction;
the feature extraction is based on LDA multi-parameter feature extraction, and LDA is used for carrying out feature extraction on the smoothed data SI' so as to achieve the purpose of dimension reduction; firstly, determining an original central point of data as m, a distance between a data point and a classification center as D, and setting a data classification number as n; the LDA is utilized to extract fault characteristics of a plurality of acquired track parameters, the acquired track parameters are characterized by being lambda (lambda 1, lambda 2, … and lambda N), N represents the number of the parameters, and further preparation is made for later-stage characteristic fusion; the method comprises the following specific steps:
calculating the mean value of the t-th sample
Figure FDA0003722332980000012
ri represents the number of samples belonging to the category t, and x represents a sample;
calculating an overall sample mean
Figure FDA0003722332980000013
n represents the number of categories, xi represents the ith sample;
defining an inter-class dispersion matrix and an intra-class dispersion matrix:
Figure FDA0003722332980000014
Figure FDA0003722332980000015
introducing Fisher discrimination criterion expression
Figure FDA0003722332980000016
The feature fusion is multi-parameter weighted fusion of genetic algorithm optimization weight coefficients, a Genetic Algorithm (GA) is used for optimizing a feature lambda fusion weight coefficient mu (mu 1, mu 2, …, mu N) to obtain an optimal weight coefficient mu OPT (mu OPT1, mu OPT2, … mu OPTN), and further the optimal fault feature fusion is achieved to maximize the obtained fusion feature fault diagnosis precision, and the fusion feature fault diagnosis precision serves as a later-stage fault diagnosis premise;
setting the genetic number as M, the genetic algebra as L, the cross probability as px, the mutation probability as pm, the binary number of the variable as PR, the surrogate furrow as GGA and the objective function as f;
the feature fusion calculation is as follows:
Figure FDA0003722332980000021
the objective function is set as follows:
max (PRE), PRE is referred to as classification accuracy;
the fault diagnosis and early warning are based on the fault diagnosis and early warning of a BP neural network, the number of input layer nodes is set to be I, the number of output layers is set to be O, and the number of hidden layer nodes is set to be H;
firstly, training BP by using a training set, and learning to obtain data characteristics; secondly, testing by using the test set, and alarming if the diagnosed fault reaches a specified level;
therefore, the number of hidden nodes is as follows:
Figure FDA0003722332980000022
the network function is defined as follows:
Figure FDA0003722332980000023
Q＝(q 1 ，…，q V ) Representing the weight coefficient, b representing the bias;
the error function is defined as:
Figure FDA0003722332980000024
the weight and bias updating formula of each iteration is as follows:
Figure FDA0003722332980000025
Figure FDA0003722332980000026
α is the learning rate, and its value range is (0, 1).
CN201810312659.9A 2018-04-09 2018-04-09 Multi-parameter fusion stacker track fault detection and early warning method Active CN110363208B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810312659.9A CN110363208B (en) 2018-04-09 2018-04-09 Multi-parameter fusion stacker track fault detection and early warning method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810312659.9A CN110363208B (en) 2018-04-09 2018-04-09 Multi-parameter fusion stacker track fault detection and early warning method

Publications (2)

Publication Number Publication Date
CN110363208A CN110363208A (en) 2019-10-22
CN110363208B true CN110363208B (en) 2022-08-12

Family

ID=68212198

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810312659.9A Active CN110363208B (en) 2018-04-09 2018-04-09 Multi-parameter fusion stacker track fault detection and early warning method

Country Status (1)

Country Link
CN (1) CN110363208B (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107590506A (en) * 2017-08-17 2018-01-16 北京航空航天大学 A kind of complex device method for diagnosing faults of feature based processing

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107590506A (en) * 2017-08-17 2018-01-16 北京航空航天大学 A kind of complex device method for diagnosing faults of feature based processing

Also Published As

Publication number Publication date
CN110363208A (en) 2019-10-22

Similar Documents

Publication Publication Date Title
CN113723010B (en) Bridge damage early warning method based on LSTM temperature-displacement correlation model
CN108875771B (en) Fault classification model and method based on sparse Gaussian Bernoulli limited Boltzmann machine and recurrent neural network
CN111562108A (en) Rolling bearing intelligent fault diagnosis method based on CNN and FCMC
CN107516135A (en) A kind of automation monitoring learning method for supporting multi-source data
CN111813084A (en) Mechanical equipment fault diagnosis method based on deep learning
CN113486578A (en) Method for predicting residual life of equipment in industrial process
CN115496163A (en) Construction abnormal energy consumption detection method based on K-Means + +, BP neural network and isolated forest
CN113157732A (en) Underground scraper fault diagnosis method based on PSO-BP neural network
CN115470962A (en) LightGBM-based enterprise confidence loss risk prediction model construction method
CN113361202A (en) Distribution transformer heavy overload rapid early warning method based on characteristic load prediction
CN115640526A (en) Drilling risk identification model, building method, identification method and computer equipment
CN112819087B (en) Method for detecting abnormality of BOD sensor of outlet water based on modularized neural network
CN110363208B (en) Multi-parameter fusion stacker track fault detection and early warning method
CN109635008B (en) Equipment fault detection method based on machine learning
CN117171713A (en) Cross self-adaptive deep migration learning method and system based on bearing service life
CN116720095A (en) Electrical characteristic signal clustering method for optimizing fuzzy C-means based on genetic algorithm
CN117009900A (en) Internet of things signal anomaly detection method and system based on graph neural network
CN116304941A (en) Ocean data quality control method and device based on multi-model combination
CN115758259A (en) Multi-source domain generalization equipment fault diagnosis method based on feature fusion
CN114372181B (en) Equipment production intelligent planning method based on multi-mode data
CN113076982B (en) Fault diagnosis and test method based on proportional valve shaft controller
CN113033683B (en) Industrial system working condition monitoring method and system based on static and dynamic joint analysis
CN113657520A (en) Intrusion detection method based on deep confidence network and long-time and short-time memory network
CN115643104A (en) Network intrusion detection method based on deep supervision discrete hash
CN115659323A (en) Intrusion detection method based on information entropy theory and convolution neural network

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