CN106597847B - Maneuvering load controller based on recurrent neural network and control method thereof - Google Patents

Maneuvering load controller based on recurrent neural network and control method thereof Download PDF

Info

Publication number
CN106597847B
CN106597847B CN201610985231.1A CN201610985231A CN106597847B CN 106597847 B CN106597847 B CN 106597847B CN 201610985231 A CN201610985231 A CN 201610985231A CN 106597847 B CN106597847 B CN 106597847B
Authority
CN
China
Prior art keywords
neural network
identification
recurrent
network
output
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
CN201610985231.1A
Other languages
Chinese (zh)
Other versions
CN106597847A (en
Inventor
黄锐
李鸿坤
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.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
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 Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN201610985231.1A priority Critical patent/CN106597847B/en
Publication of CN106597847A publication Critical patent/CN106597847A/en
Application granted granted Critical
Publication of CN106597847B publication Critical patent/CN106597847B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Abstract

The invention discloses a maneuvering load controller based on a recurrent neural network, wherein the neural network of the maneuvering load controller is formed by connecting a recurrent identification neural network and a recurrent control neural network in series, and 6 neurons are adopted to form a network respectively; the invention also discloses a control method of the maneuvering load controller based on the recurrent neural network; the method using the recurrent neural network has the characteristics of simple structure, good working performance stability, high robustness and good self-adaption, and can effectively carry out self-adaption maneuvering load alleviation on the airplane flying at the variable Mach number; the invention realizes the design of the maneuvering load controller of the airplane based on the recurrent neural network in the wide Mach number range and realizes the self-adaptive control of the maneuvering load of the airplane under the complex flight condition.

Description

Maneuvering load controller based on recurrent neural network and control method thereof
Technical Field
The invention belongs to the technical field of aircraft control, and particularly relates to a maneuvering load controller based on a recurrent neural network and a control method thereof.
Background
When the airplane does maneuvering action, a large bending moment is generated at the wing root, so that the service life of the airplane is influenced, and potential safety hazards are brought. The aerodynamic force of the airplane wing can be redistributed by using the maneuvering load controller, and the aim of reducing the bending moment of the wing root is achieved.
Traditional maneuvering load controllers, such as PID (proportion integration differentiation) can not realize adaptive control of variable Mach number, when flight conditions change, parameters of the controllers need to be selected again, real-time online control can not be achieved, and effective effect can not be achieved in the actual flight process of an airplane. Therefore, designing a maneuvering load controller and using the controller to realize the airplane maneuvering load alleviation has been a technical problem to be solved by those skilled in the art.
Disclosure of Invention
The invention discloses a maneuvering load controller based on a recurrent neural network and a control method thereof, aiming at the problems in the prior art, the recurrent neural network is used for carrying out two stages of system identification and maneuvering load reduction, the reduction of wing root bending moment of an airplane during variable Mach number flight can be realized, and the maneuvering action of the airplane can be ensured to be normally finished.
The invention is realized in such a way that the invention discloses a maneuvering load controller based on a recurrent neural network, the neural network of the maneuvering load controller is formed by connecting a recurrent identification neural network and a recurrent control neural network in series, and 6 neurons are respectively adopted to form a network.
Further, the basic mathematical model of the neural network is as follows:
Figure GDA0001213569210000011
Figure GDA0001213569210000012
in the formula, vkFor internal net activation, ujRepresenting an input vector, wkjA weight matrix representing the synapses,
Figure GDA0001213569210000013
to activate a function, ykRepresenting the output vector, p is the total number of neural network inputs, including the bias.
The invention also discloses a control method of the maneuvering load controller based on the recurrent neural network, which comprises the following steps:
step 1, taking a normal overload signal, an attack angle and a wing root bending moment signal of an airplane as input, and taking deflection of a horizontal tail and an aileron as output; the normal overload and the attack angle of the normal overload aircraft can be collected through corresponding sensors, the wing root bending moment can be obtained through strain signals in a mode of attaching strain gauges to the wing roots, and the wing root bending moment can be calculated in real time. The deflection of the horizontal tail and the aileron is realized by an actuator.
Step 2, identifying the system by adjusting parameters of a recurrent identification neural network activation function and an initial weight matrix;
the recursive identification network is used to identify the sensor signals of the system. At time n, recursively identifying network prefixesMeasuring the signal at time n +1
Figure GDA0001213569210000021
As an input to a control network; the recursion identification neural network carries out online learning through a real-time recursion algorithm and identifies normal overload, an attack angle and a wing root bending moment signal acquired by the sensor.
And 3, normal overload, tracking of an attack angle and suppression of a wing root bending moment are performed by adjusting parameters of a recursive control neural network activation function and an initial weight matrix. The recursion control neural network also adopts a real-time recursion algorithm to carry out online learning, and normal overload, an attack angle and a wing root bending moment which are output by the recursion identification neural network are taken as input signals, so that the real-time tracking of the normal overload and the attack angle and the real-time control of the wing root bending moment are realized.
Further, the activation functions adopted by the recurrent identification neural network and the recurrent control neural network are as follows:
Figure GDA0001213569210000022
neural network static activation vjAnd input uiSatisfies the following conditions:
Figure GDA0001213569210000023
in the formula: w is aijIs a weight matrix; v. ofjFor static activation of neural networks, ujA vector is input to the neural network.
Further, the step 2 specifically comprises:
2.1, the recursion identification network carries out online learning through a real-time recursion algorithm, and calculates an error gradient, wherein an error function of the recursion identification neural network is as follows:
Figure GDA0001213569210000024
in the formula, EidAs an objective function, nm is the total number of sensors,
Figure GDA0001213569210000025
is the error between the actual sensor signal and the identification signal.
2.2, according to the error between the actual sensor signal and the identification signal, identifying the nth time step in the recursion identification neural network, wherein the expression is as follows:
Figure GDA0001213569210000026
the output of the neural network neurons of the recursive recognition is:
Figure GDA0001213569210000027
Figure GDA0001213569210000031
in the formula (I), the compound is shown in the specification,
Figure GDA0001213569210000032
is the output of the neuron, and nid is the total number of inputs of the neural network.
2.3 identifying the design parameters of the network recursively by means of an identification process, i.e. EidThe method achieves the minimum, and comprises the following specific processes:
Figure GDA0001213569210000033
Figure GDA0001213569210000034
in the formula, ηidIs the learning rate of the neural network.
Further, the step 3 specifically comprises:
3.1, the recursion control neural network is similar to the recursion identification neural network in structure, and the output expression of the recursion control neural network neuron is as follows:
Figure GDA0001213569210000035
the objective function is:
Figure GDA0001213569210000036
the error between the actual output and the expected output of the system is expressed as:
Figure GDA0001213569210000037
the minimization of the objective function requires iteration of the weight matrix:
Figure GDA0001213569210000038
the objective function gradient can be expressed as:
Figure GDA0001213569210000039
wherein nm is the total output number of the recursive identification network; ni is the total number of outputs of the recursive control network, and γ is a parameter that regulates net activation of the controller; e.g. of the typecoη as an error between the actual output and the desired output of the systemcoThe learning rate of the network is controlled recursively. The recursive control neural network still uses a real-time recursive algorithm for online learning.
Compared with the prior art, the invention has the beneficial effects that: the maneuvering load controller based on the recurrent neural network has self-adaptability and can realize maneuvering load alleviation of the airplane flying at variable Mach number; the method using the recurrent neural network has the characteristics of simple structure, good working performance stability, high robustness and good self-adaption, and can effectively carry out self-adaption maneuvering load alleviation on the airplane flying at the variable Mach number; the invention realizes the design of the maneuvering load controller of the airplane based on the recurrent neural network in the wide Mach number range and realizes the self-adaptive control of the maneuvering load of the airplane under the complex flight condition.
Drawings
FIG. 1 is a schematic structural diagram of a motor load controller based on a recurrent neural network according to the present invention;
FIG. 2 is a schematic illustration of the present invention for achieving aircraft maneuvering load mitigation;
FIG. 3 is a schematic diagram of control surface deflection during the system identification stage according to the present invention;
FIG. 4 is a schematic illustration of the horizontal tail deflection during a maneuver load alleviation phase of the present invention;
FIG. 5 is a diagram illustrating a system identification result according to the present invention;
FIG. 6 is a graph illustrating the load alleviation results of the present invention;
FIG. 7 is a schematic illustration of the variable Mach number load mitigation of the present invention;
fig. 8 is a schematic view of the closed loop control surface deflection of the present invention.
Detailed Description
The invention provides a maneuvering load controller based on a recurrent neural network and a control method thereof, and in order to make the purpose, technical scheme and effect of the invention clearer and clearer, the invention is further described in detail by referring to the attached drawings and taking examples. It should be noted that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Fig. 1 is a schematic structural diagram of a motor load controller based on a recurrent neural network. The controller is formed by connecting a recursion identification neural network and a recursion control neural network, 6 neurons are adopted to form the network respectively, and the basic mathematical model of the neural network is as follows:
Figure GDA0001213569210000041
Figure GDA0001213569210000042
the recursive identification network is used to identify the sensor signals of the system. At time n, the recursive recognition network predicts the signal at time n +1
Figure GDA0001213569210000043
As input to the control network.
The recursive identification network performs online learning through a real-time recursive algorithm. The real-time recursive algorithm is a gradient descent method, and the main algorithm is to calculate the error gradient of a specific expression at each time step. The error function for recursively identifying the neural network is:
Figure GDA0001213569210000051
the expression of the error between the actual sensor signal and the identification signal at the nth time step for identifying the recurrent identification neural network is as follows:
Figure GDA0001213569210000052
the output of the neural network neurons of the recursive recognition is:
Figure GDA0001213569210000053
Figure GDA0001213569210000054
the goal of the identification process is to adjust the weight matrix by each time step so that the design parameter of the network, E, is recursively identifiedidThe method achieves the minimum, and comprises the following specific processes:
Figure GDA0001213569210000055
Figure GDA0001213569210000056
the recursion control neural network is similar to the recursion identification neural network in structure, and the output expression of the recursion control neural network neuron is as follows:
Figure GDA0001213569210000057
the objective function is:
Figure GDA0001213569210000058
the error between the actual output and the expected output of the system is expressed as:
Figure GDA0001213569210000059
minimization of the objective function requires iteration of the weight matrix:
Figure GDA00012135692100000510
the objective function gradient can be expressed as:
Figure GDA0001213569210000061
the recursion control neural network still uses a real-time recursion algorithm for online learning, and the specific derivation process is as follows:
Figure GDA0001213569210000062
Figure GDA0001213569210000063
Figure GDA0001213569210000064
Figure GDA0001213569210000065
Figure GDA0001213569210000066
Figure GDA0001213569210000067
Figure GDA0001213569210000068
Figure GDA0001213569210000069
Figure GDA00012135692100000610
Figure GDA00012135692100000611
where nco is the total number of recursively controlled network inputs, δtrIs a kronecker symbol and nhi is the number of implicit inputs, i.e., the number of virtual outputs fed back by the recursive control network.
Fig. 2 is a schematic diagram of the motor load controller based on the recurrent neural network to realize the reduction of the motor load of the airplane, the input of the controller comprises the normal overload of the airplane, the flight angle of attack of the airplane and the wing root bending moment, and the output comprises the horizontal tail deflection angle and the aileron deflection angle. The maneuvering load controller based on the recurrent neural network of the invention realizes the slowing of the maneuvering load of the airplane into two stages: 1. system identification; 2. the maneuvering load is slowed down. Specific examples are as follows:
the controller parameters were designed at mach number 0.6.
The system identification phase lasts 10s, the initial inputs of the horizontal tail and the aileron are shown in FIG. 3, and the parameters of the neural network are identified recursively as follows:
learning rate of neural network ηid＝0.5
The activation function is:
Figure GDA0001213569210000071
the optimal initial weight matrix of the recurrent identification neural network is obtained through numerical simulation. The system identification result is shown in fig. 5, and it can be seen that the output of the recursive identification network can quickly and accurately track the open-loop output of the system.
The maneuver load alleviation phase lasts 10 seconds and the open loop input deflects only the horizontal tail as shown in FIG. 4. The parameters of the neural network are recursively controlled as follows:
learning rate of neural network ηco＝0.02
The activation function is:
Figure GDA0001213569210000072
the optimal initial weight matrix of the recurrent control neural network is obtained through numerical simulation. The system control results are shown in fig. 6. As can be seen from the figure, at mach number 0.6, the maneuvering load controller based on the recurrent neural network can reduce the wing root bending moment ratio by 26% without using the controller, and the normal overload and the attack angle are consistent with those without using the controller.
After the parameter design of the maneuvering load controller based on the recurrent neural network is finished, the parameter is applied to other Mach numbers, and a parameter lambda related to the Mach number is introduced to be used as a multiplier of the recurrent identification neural network initial weight matrix and the recurrent control neural network initial weight matrix, wherein the relation between the lambda and the Mach number Ma is as follows:
λ＝0.29352-1.969251Ma+10.0211Ma2-8.10263Ma3
the parameter is suitable for the airplane to do maneuvering action within the range of Mach number 0.3 to Mach number 0.69, the maneuvering load reducing situation under various Mach numbers is shown in FIG. 7, and the maximum wing root bending moment can be reduced by 49%. The flattail and aileron deflections at different mach numbers are shown in figure 8.
The foregoing is a preferred embodiment of the invention, and the description and use of the invention herein is illustrative; it should be noted that modifications can be made by those skilled in the art without departing from the principle of the present invention, and these modifications should also be construed as the scope of the present invention.

Claims (3)

1. A control method of a maneuvering load controller based on a recurrent neural network is characterized by comprising the following specific steps:
step 1, taking a normal overload signal, an attack angle and a wing root bending moment signal of an airplane as input, and taking deflection of a horizontal tail and an aileron as output;
step 2, identifying the system by adjusting parameters of the recurrent identification neural network activation function and the initial weight matrix, specifically:
2.1, the recursion identification network carries out online learning through a real-time recursion algorithm, and calculates an error gradient, wherein an error function of the recursion identification neural network is as follows:
Figure FDA0002290950880000011
in the formula, EidAs an objective function, nm is the total number of sensors,
Figure FDA0002290950880000012
is the error of the actual sensor signal and the identification signal;
2.2, according to the error between the actual sensor signal and the identification signal, identifying the nth time step in the recursion identification neural network, wherein the expression is as follows:
Figure FDA0002290950880000013
the output of the neural network neurons of the recursive recognition is:
Figure FDA0002290950880000014
Figure FDA0002290950880000015
in the formula (I), the compound is shown in the specification,
Figure FDA0002290950880000018
is the output of the neuron, and nid is the total input number of the neural network;
2.3 identifying the design parameters of the network recursively by means of an identification process, i.e. EidThe method achieves the minimum, and comprises the following specific processes:
Figure FDA0002290950880000016
Figure FDA0002290950880000017
in the formula, ηidLearning rate for neural networks;
and 3, normal overload, tracking of an attack angle and suppression of a wing root bending moment are performed by adjusting parameters of a recursive control neural network activation function and an initial weight matrix.
2. The method as claimed in claim 1, wherein the activation functions used by the recurrent identification neural network and the recurrent control neural network are:
Figure FDA0002290950880000021
neural network static activation vjAnd input uiSatisfies the following conditions:
Figure FDA0002290950880000022
in the formula: w is aijIs a weight matrix, vjFor static activation of neural networks, ui(n) is the input to the neuron at the nth time step.
3. The method for controlling a maneuvering load controller based on a recurrent neural network as claimed in claim 1, characterized in that said step 3 is specifically:
the recursion control neural network uses a real-time recursion algorithm to carry out online learning, and the output expression of the recursion control neural network neurons is as follows:
Figure FDA0002290950880000023
the objective function is:
Figure FDA0002290950880000024
the error between the actual output and the expected output of the system is expressed as:
Figure FDA0002290950880000025
iterating the weight matrix:
Figure FDA0002290950880000026
the objective function gradient can be expressed as:
Figure FDA0002290950880000031
wherein nm is the total output number of the recursive identification network; ni is the total number of outputs of the recursive control network, and γ is a parameter that regulates net activation of the controller; e.g. of the typecoη as an error between the actual output and the desired output of the systemcoThe learning rate of the network is controlled recursively.
CN201610985231.1A 2016-11-09 2016-11-09 Maneuvering load controller based on recurrent neural network and control method thereof Active CN106597847B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610985231.1A CN106597847B (en) 2016-11-09 2016-11-09 Maneuvering load controller based on recurrent neural network and control method thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610985231.1A CN106597847B (en) 2016-11-09 2016-11-09 Maneuvering load controller based on recurrent neural network and control method thereof

Publications (2)

Publication Number Publication Date
CN106597847A CN106597847A (en) 2017-04-26
CN106597847B true CN106597847B (en) 2020-03-17

Family

ID=58589887

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610985231.1A Active CN106597847B (en) 2016-11-09 2016-11-09 Maneuvering load controller based on recurrent neural network and control method thereof

Country Status (1)

Country Link
CN (1) CN106597847B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109143856A (en) * 2018-07-31 2019-01-04 佛山科学技术学院 Adaptive health indicator extracting method based on depth recurrent neural network
CN110703603B (en) * 2019-10-28 2020-11-24 华南理工大学 Control method of multi-layer recursive convergence neural network controller of unmanned aerial vehicle

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6125333A (en) * 1997-11-06 2000-09-26 Northrop Grumman Corporation Building block approach for fatigue spectra generation
CN104390776A (en) * 2014-12-10 2015-03-04 北京航空航天大学 Fault detection, diagnosis and performance evaluation method for redundant aileron actuator
CN104834808A (en) * 2015-04-07 2015-08-12 青岛科技大学 Back propagation (BP) neural network based method for predicting service life of rubber absorber
CN105488281A (en) * 2015-12-01 2016-04-13 北京航空航天大学 Method for identifying airplane structure load based on flight parameter monitoring
CN205300843U (en) * 2015-12-26 2016-06-08 哈尔滨佳云科技有限公司 Civilian helicopter moves loading test device

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8024079B2 (en) * 2008-05-20 2011-09-20 The Boeing Company Wing-body load alleviation for aircraft
US10017243B2 (en) * 2013-09-24 2018-07-10 The Boeing Company Adaptive trailing edge actuator system and method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6125333A (en) * 1997-11-06 2000-09-26 Northrop Grumman Corporation Building block approach for fatigue spectra generation
CN104390776A (en) * 2014-12-10 2015-03-04 北京航空航天大学 Fault detection, diagnosis and performance evaluation method for redundant aileron actuator
CN104834808A (en) * 2015-04-07 2015-08-12 青岛科技大学 Back propagation (BP) neural network based method for predicting service life of rubber absorber
CN105488281A (en) * 2015-12-01 2016-04-13 北京航空航天大学 Method for identifying airplane structure load based on flight parameter monitoring
CN205300843U (en) * 2015-12-26 2016-06-08 哈尔滨佳云科技有限公司 Civilian helicopter moves loading test device

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
一种递归模糊神经网络自适应控制方法;毛六平,等;《电子学报》;20061231;第34卷(第12期);第2285-2287页 *
刚弹耦合飞行器的机动载荷减缓;唐皓，等;《航空计算技术》;20120531;第42卷(第3期);第33-37页 *

Also Published As

Publication number Publication date
CN106597847A (en) 2017-04-26

Similar Documents

Publication Publication Date Title
Xu et al. Neural network based dynamic surface control of hypersonic flight dynamics using small-gain theorem
CN108803336B (en) Design method of self-adaptive LQG/LTR controller of aircraft engine
CN110161841A (en) A kind of feedforward-fuzzy PID control method suitable for temporarily rushing formula transonic wind tunnel
Wu et al. Adaptive terminal sliding mode control for hypersonic flight vehicles with strictly lower convex function based nonlinear disturbance observer
Khalid et al. Conventional PID, adaptive PID, and sliding mode controllers design for aircraft pitch control
CN106597847B (en) Maneuvering load controller based on recurrent neural network and control method thereof
CN111474922B (en) Method for constructing controller of continuous nonlinear system
CN112286047A (en) NARMA-L2 multivariable control method based on neural network
CN111486009A (en) Aero-engine control method and device based on deep reinforcement learning
Wu et al. Adaptive neural flight control for an aircraft with time-varying distributed delays
Lazim et al. Intelligent observer-based feedback linearization for autonomous quadrotor control
CN110985216B (en) Intelligent multivariable control method for aero-engine with online correction
CN115685764B (en) Task self-adaptive anti-interference tracking control method and system for variable-span aircraft
Moin et al. State space model of an aircraft using Simulink
CN114003052B (en) Fixed wing unmanned aerial vehicle longitudinal movement robust self-adaptive control method based on dynamic compensation system
CN102707629A (en) Design method of full-dimensional controller region based on aircraft switching model
CN103809434B (en) The multistage PID controller design method of the compound root locus of Longitudinal Flight model cluster
CN114237268A (en) Unmanned aerial vehicle strong robust attitude control method based on deep reinforcement learning
Tang et al. A hybrid control combining an advanced sliding mode control law with a proportional-derivative control by applying on vibration suppression
Lv et al. An adaptive dynamic surface controller for ultralow altitude airdrop flight path angle with actuator input nonlinearity
Nguyen Robust optimal adaptive control method with large adaptive gain
Păucă et al. Comparative study on pitch angle control of an aircraft using neural networks
CN109799765A (en) Steering engine electrohydraulic servo system intelligent control method
CN108762070A (en) A kind of fractional order control method of drive lacking unmanned plane
Qian et al. Modeling and Fitting of Aircraft Fly-by-wire Control System Based on Improved BP 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