CN113218520B - Optimized neural network extraction method for laser pulse width - Google Patents
Optimized neural network extraction method for laser pulse width Download PDFInfo
- Publication number
- CN113218520B CN113218520B CN202110478101.XA CN202110478101A CN113218520B CN 113218520 B CN113218520 B CN 113218520B CN 202110478101 A CN202110478101 A CN 202110478101A CN 113218520 B CN113218520 B CN 113218520B
- Authority
- CN
- China
- Prior art keywords
- signal
- neural network
- time
- sequence
- time sequence
- 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
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 15
- 238000000605 extraction Methods 0.000 title abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000003062 neural network model Methods 0.000 claims abstract description 29
- 238000005259 measurement Methods 0.000 claims abstract description 10
- 230000003287 optical effect Effects 0.000 claims abstract description 9
- 238000004458 analytical method Methods 0.000 claims abstract description 7
- 230000000903 blocking effect Effects 0.000 claims abstract description 7
- 238000010521 absorption reaction Methods 0.000 claims abstract description 6
- 230000000694 effects Effects 0.000 claims abstract description 6
- 239000004065 semiconductor Substances 0.000 claims abstract description 6
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000000034 method Methods 0.000 claims description 18
- 210000002569 neuron Anatomy 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 208000025174 PANDAS Diseases 0.000 abstract description 3
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 abstract description 3
- 235000016496 Panda oleosa Nutrition 0.000 abstract description 3
- 240000000220 Panda oleosa Species 0.000 abstract 1
- JBRZTFJDHDCESZ-UHFFFAOYSA-N AsGa Chemical compound [As]#[Ga] JBRZTFJDHDCESZ-UHFFFAOYSA-N 0.000 description 4
- 229910001218 Gallium arsenide Inorganic materials 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 240000004718 Panda Species 0.000 description 2
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 229910052710 silicon Inorganic materials 0.000 description 2
- 239000010703 silicon Substances 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000005311 autocorrelation function Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J11/00—Measuring the characteristics of individual optical pulses or of optical pulse trains
-
- 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
-
- 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 an optimized neural network extraction method for laser pulse width, and belongs to the technical field of optical measurement. The technical scheme of the invention is as follows: carrying out intensity autocorrelation measurement by using a non-balanced Michelson interference optical path and a semiconductor two-photon absorption effect to obtain a photoelectric signal; separating a time sequence t and a signal value sequence s from the acquired intensity autocorrelation signal with the electrical noise; constructing an optimized neural network model by using a third-party deep learning toolkit (Numpy, Pandas, Matplotlib), taking a time sequence t as an input of model training, taking I as a label value of the model training, wherein intensity autocorrelation signal data is a binary group (t, u) which respectively represents the time sequence and a signal value sequence, and S is (t, I) which is marked as the ith pair of time and signal values in the intensity autocorrelation signal and represents the deviation (ith time) between a predicted signal and real signal data; and (4) performing blocking treatment on the I' to obtain a pure alternating current signal I ", and performing half-height width analysis on the pure alternating current signal I".
Description
Technical Field
The invention relates to an optimized neural network extraction method for laser pulse width, and belongs to the technical field of optical measurement.
Background
The measurement of the laser pulse width typically employs well-known optical autocorrelation techniques. Optical autocorrelators are typically constructed based on optical nonlinear effects. Recently, two-photon absorption (TPA) effects in semiconductors have been used to measure optical autocorrelation due to sensitivity and potentially low cost similar to Second Harmonic Generation (SHG) crystal autocorrelators. Research has shown that silicon (Si) and gallium arsenide (GaAs) photodiodes can be used as TPA devices in the infrared band for autocorrelation measurement of laser pulses in the communication band. The difference between Si and GaAs is mainly the measurement of the dispersed pulses. Gallium arsenide is a direct bandgap semiconductor material compared to silicon, and its electronic energy level transition does not involve simultaneous emission or absorption of phonons, which can avoid the influence of environment. It has been experimentally observed that TPA effect is more intense in gaas photodiodes. Therefore, GaAs photodiodes are superior to Si for autocorrelation measurements.
Since neither intensity nor interferometric autocorrelation fully captures the pulse information (intensity and phase), they are insufficient to determine the temporal distribution of the pulse, but we can curve fit the autocorrelation signal to measure the pulse width. For conventional fitting methods, such as the Levenberg-Marquardt method (L-M), we need to preset the pulse function. Due to the influence of electronic noise, the measured autocorrelation signal does not reach the ideal gaussian or hyperbolic secant shape, which may result in a failure of the fitting. In recent years, with the development of deep learning, Neural Networks (NN) have been increasingly used in the field of artificial intelligence. The universal approximation theorem indicates that, in theory, any function can be approximated by a neural network having at least one hidden layer. This shows its unique advantage in curve fitting.
Disclosure of Invention
The invention aims to provide a method for extracting the optimized neural network of the laser pulse width aiming at the defects of the prior art, the method is used for fitting a function suitable for pulse width analysis based on the neural network, the intensity autocorrelation of femtosecond laser is generated by utilizing the two-photon absorption effect of a semiconductor, and then the neural network fitting is carried out on the intensity autocorrelation signal so as to analyze the signal and extract the pulse width.
The technical scheme adopted for solving the technical problem comprises the following steps: a method for extracting an optimized neural network of laser pulse width comprises the following steps:
step 1: and carrying out intensity autocorrelation measurement by using an unbalanced Michelson interference optical path and a semiconductor two-photon absorption effect to obtain a photoelectric signal.
Step 2: separating the acquired intensity autocorrelation signal with the electrical noise into a time sequence t and a signal value sequence s.
And step 3: and (3) constructing an optimized neural network model by using a third-party deep learning toolkit (Numpy, Pandas, Matplotlib), taking the time sequence t as the input of model training, and taking I as the label value of the model training.
The intensity autocorrelation signal data is a two-tuple (t, u) representing a time series and a signal value series, respectively,
(t, I) as the ith pair of time and signal values in the intensity autocorrelation signal;
representing the deviation (i-th time) between the predicted signal and the true signal data.
And 4, step 4: and performing DC blocking treatment on the I 'to obtain a pure alternating current signal I'. And then carrying out half-height-width analysis on the pulse width, and finally, accurately extracting the pulse width by utilizing a zero setting algorithm.
The optimized neural network model constructed in the step 3 comprises the following steps:
step 3-1: setting the neural network model parameters, wherein the input layer and the output layer are both neurons, and the number of the hidden layers and the number of the neurons in each layer are specifically set according to the actual condition of a user. And setting the iteration times Epoch and the Performance parameter Performance, and controlling the termination condition of the model training by using whether the threshold value is reached. Then, taking the time sequence t as the input of neural network model training, and taking I as the label value of model training to perform supervised training on the model;
step 3-2: defining a random restart function in the neural network model and setting parameters. In the neural network model training process of step 3-1, when the iteration number exceeds a certain value, each parameter is reinitialized, and the training is carried out again.
Step 3-3: a new time series t' of high sampling rates is generated, the time range covering exactly one complete pulse. And (5) sending the time sequence t 'into the generated neural network model to obtain a new signal value sequence I'.
Furthermore, the increment of the time sequence t 'generated each time is different from the increment of the time sequence t of the acquired real signal, and the total length of the time sequence t' is less than the length of the time sequence t of the real signal; meanwhile, the method removes the strong direct current value by using the characteristic that the strong direct current value exists in the generated strength autocorrelation signal and using a blocking algorithm and a zero setting algorithm, thereby obtaining the signal which is easy to extract the pulse width.
Further, the invention has a user adjustment function, and can adjust the random restart condition aiming at different autocorrelation signals.
Furthermore, the invention can determine the prediction resolution of the neural network model by adjusting the iteration number Epoch and the Performance parameter Performance, thereby influencing the extraction accuracy of the pulse width.
Furthermore, the user can customize the number of hidden layers and the number of neurons in each layer according to actual requirements, and select the number of layers and the number of neurons with the shortest training time and the optimal precision.
Has the advantages that:
1. the method utilizes the neural network model to carry out accurate function generation on the strength autocorrelation signal with the electrical noise, generates a new autocorrelation function which is easy to carry out pulse width analysis in real time, has the characteristic of noise resistance, and avoids the problem that the traditional pulse fitting is easy to fail in fitting the noise-containing signal.
2. The invention adopts the random restart algorithm to optimize the neural network model and improves the defect that the neural network is easy to fall into the local optimum.
3. The model generated by the invention can be matched with any complex autocorrelation signal, and the intensity autocorrelation frequency at any moment can be well extracted by using a time-frequency analysis method.
4. The method has the advantages of simple model training process and low calculation cost, and can ensure that the result of frequency domain extraction is credible rather than disordered.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
In order to more clearly illustrate the technical solutions of the present invention, the following description is given with reference to specific embodiments and accompanying drawings, and it is obvious that the embodiments in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other embodiments can be obtained according to these embodiments without any inventive work.
As shown in fig. 1, the method for measuring a laser pulse width based on a neural network according to the present invention includes the following steps:
step 1: determining a target, and determining the target as pulse width information extraction according to requirements;
step 2: data acquisition and integration, wherein the intensity autocorrelation signal data required by the realization of the target are acquired and integrated;
and step 3: data cleaning, namely separating a time sequence and a signal value sequence of the acquired data;
and 4, step 4: selecting and constructing a neural network model, and selecting, constructing and optimizing parameters of the neural network model according to the requirement and the target of pulse width measurement;
and 5: training and evaluating the model, training and evaluating the optimized neural network to generate a new strength autocorrelation signal;
step 6: and (4) extracting the pulse width, namely specifically extracting the required pulse width by using a zero setting method.
To facilitate the realization of the concepts of the invention by those skilled in the art, a specific embodiment is now provided, including the following:
step 1: determining a target, and determining required target parameters according to the requirements of a user;
step 2: data collection and integration, wherein relevant strength autocorrelation signal data of a specific target are collected and integrated;
and step 3: data cleaning, namely separating a time sequence and a signal value sequence from the acquired data, namely separating the time sequence t and the signal value sequence s from the acquired intensity autocorrelation signal;
and 4, step 4: selecting and constructing a neural network model, selecting and constructing parameters of the neural network model according to the requirements and targets of users, wherein the constructed neural network model takes a time sequence t as the input of model training, I as the label value of the model training, the strength autocorrelation signal data is a binary group (t, I) which respectively represents the time sequence and the signal value sequence,
S＝(t，I),Sidenoted as the ith pair of time and signal values in the intensity autocorrelation signal,
ΔSirepresents the deviation between the predicted signal and the true signal data (i-th time);
and 5: model training and evaluation, namely evaluating the obtained model to generate a new strength autocorrelation signal;
step 6: analyzing the pulse width information, and specifically extracting the required pulse width information by using a zero setting algorithm. And performing blocking processing on the generated new signal value sequence to obtain a pure alternating current signal, performing zero setting processing on the pure alternating current signal to obtain a square wave signal, and finally, accurately extracting the pulse width by using a time extraction algorithm.
The construction of the optimized neural network model comprises the following steps:
step 4-1: the neural network model parameters are set, the input layer and the output layer are both neurons, and the number of the hidden layers and the number of the neurons in each layer are specifically set according to actual conditions. And setting the iteration times Epoch and the Performance parameter Performance, and controlling the termination condition of the model training by using whether the threshold value is reached. Then, taking the time sequence t as the input of model training, and taking I as the label value of neural network model training to perform supervised training on the model;
step 4-2: a random restart function is defined in the model, and parameters are set. In the model training process of the step 3-1, when the iteration times exceed a certain value, all parameters are reinitialized, and training is carried out again;
step 4-3: a new time series t' of high sampling rates is generated, the time range covering exactly one complete pulse. Sending the time sequence t 'into the generated neural network model to obtain a new signal value sequence I';
the time sequence t' generated each time and the time sequence t of the acquired real signal have different interval ranges and sampling rates.
The user can determine the predicted resolution of the neural network model by adjusting the iteration number Epoch and the Performance parameter Performance, so that the pulse width extraction precision is influenced.
The user can adjust the conditions for random restart for different autocorrelation signals.
The user can customize the number of hidden layers and the number of neurons in each layer according to actual requirements.
The user can select the number of layers and the number with the shortest training time and the best precision.
In specific implementation, a third-party deep learning toolkit (Numpy, Pandas, Matplotlib) can be used for constructing a neural network model, the time sequence t is used as the input of model training, and the I is used as the label value of the model training.
The intensity autocorrelation signal data is a doublet (t, I) representing a time series and a signal value series, respectively,
S＝(t，I）,Sirecording as the ith pair of time and signal values in the intensity autocorrelation signal;
ΔSirepresents the deviation (i-th time) between the predicted signal and the true signal data, i representing the index of the time-signal value pair;
when a neural network model is constructed, the number of hidden layers and the number of neurons in each layer are specifically set according to actual conditions; while the conditions for random restart are adjusted for different autocorrelation signals.
And during blocking and pulse width analysis, blocking the new signal value sequence I 'to obtain a pure alternating current signal I', carrying out zero setting processing on the pure alternating current signal I to obtain a square wave signal, and finally accurately extracting the pulse width by using a time extraction algorithm.
The parts not involved in the present invention are the same as or can be implemented using the prior art.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (2)
1. A method for extracting a neural network for optimizing laser pulse width is characterized by comprising the following steps:
step 1: carrying out intensity autocorrelation measurement by using a non-balanced Michelson interference optical path and a semiconductor two-photon absorption effect to obtain a photoelectric signal;
step 2: separating a time sequence t and a signal value sequence I from the acquired intensity autocorrelation signal with the electrical noise;
and step 3: constructing an optimized neural network model by utilizing a Python third-party deep learning toolkit, taking a time sequence t as input of model training, taking I as a label value of the model training, wherein strength autocorrelation signal data is a binary group (t, I), and t and I respectively represent the time sequence and a signal value sequence;
S＝(t，I),Sirecording as the ith pair of time and signal values in the intensity autocorrelation signal;
ΔSirepresenting the deviation between the ith prediction signal and the true signal data;
the constructed optimized neural network model comprises the following steps:
step 3-1: setting parameters of a neural network model, setting the number of input and output layers as a neuron, specifically setting the number of layers of hidden layers and the number of neurons of each layer according to the actual condition of a user, setting the iteration times Epoch and Performance parameters, controlling the termination condition of the model training by using whether a threshold value is reached, and then carrying out supervised training on the model by using a time sequence t as the input of the model training and I as the label value of the neural network model training;
step 3-2: defining a random restart function in the neural network model, setting parameters, and reinitializing the parameters and performing training again when the iteration times exceed a certain value in the model training process in the step 3-1;
step 3-3: generating a new time sequence t ' with high sampling rate, wherein the time range just covers a complete pulse, and sending the time sequence t ' into the generated neural network model to obtain a new signal value sequence I ';
and 4, step 4: and (3) carrying out blocking processing on the new signal value sequence I ' to obtain a pure alternating current signal I ', carrying out half-height width analysis on the pure alternating current signal I ', and finally accurately extracting the pulse width by utilizing a zero setting algorithm.
2. The method as claimed in claim 1, wherein the time series t 'generated each time is different from the increment of the time series t of the real signal, and the total length of the time series t' is smaller than the length of the time series t of the real signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110478101.XA CN113218520B (en) | 2021-04-30 | 2021-04-30 | Optimized neural network extraction method for laser pulse width |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110478101.XA CN113218520B (en) | 2021-04-30 | 2021-04-30 | Optimized neural network extraction method for laser pulse width |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113218520A CN113218520A (en) | 2021-08-06 |
| CN113218520B true CN113218520B (en) | 2021-11-09 |
Family
ID=77090257
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110478101.XA Active CN113218520B (en) | 2021-04-30 | 2021-04-30 | Optimized neural network extraction method for laser pulse width |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113218520B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114444293B (en) * | 2022-01-21 | 2023-06-23 | 南京森林警察学院 | Interference-like signal phase decomposition method based on transfer learning |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0626937A (en) * | 1992-07-08 | 1994-02-04 | Advantest Corp | Measuring method of optical pulse waveform |
| CN106855437A (en) * | 2016-12-12 | 2017-06-16 | 西北大学 | A kind of single-shot ultraviolet ultrashort-pulse pulse width measure device and method |
| CN109883557A (en) * | 2019-02-28 | 2019-06-14 | 上海交通大学 | Integrated optics autocorrelation function analyzer and measurement method based on selenizing molybdenum two-photon absorption |
| CN111814578A (en) * | 2020-06-15 | 2020-10-23 | 南京森林警察学院 | Method for extracting frequency of ultralow frequency Doppler signal |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7038781B2 (en) * | 2003-10-01 | 2006-05-02 | Coherent, Inc. | Time correlation of ultrafast laser pulses |
-
2021
- 2021-04-30 CN CN202110478101.XA patent/CN113218520B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0626937A (en) * | 1992-07-08 | 1994-02-04 | Advantest Corp | Measuring method of optical pulse waveform |
| CN106855437A (en) * | 2016-12-12 | 2017-06-16 | 西北大学 | A kind of single-shot ultraviolet ultrashort-pulse pulse width measure device and method |
| CN109883557A (en) * | 2019-02-28 | 2019-06-14 | 上海交通大学 | Integrated optics autocorrelation function analyzer and measurement method based on selenizing molybdenum two-photon absorption |
| CN111814578A (en) * | 2020-06-15 | 2020-10-23 | 南京森林警察学院 | Method for extracting frequency of ultralow frequency Doppler signal |
Non-Patent Citations (1)
| Title |
|---|
| 神经网络在时间序列预测中的应用研究;徐鹏飞等;《电子技术》;20100825;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113218520A (en) | 2021-08-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107402082B (en) | Brillouin scattering signal processing method and distributed optical fiber sensing system thereof | |
| CN111442827B (en) | Optical fiber passive online monitoring system for transformer winding vibration | |
| CN106500735B (en) | A kind of compressed sensing based FBG signal adaptive restorative procedure | |
| CN103989462B (en) | The extracting method of a kind of pulse wave fisrt feature point and second feature point | |
| CN111273336B (en) | Gaussian forming method for digital nuclear pulse signal | |
| US20030115017A1 (en) | Method and apparatus for analyzing a disribution | |
| CN103941254A (en) | Soil physical property classification recognition method and device based on geological radar | |
| CN113218520B (en) | Optimized neural network extraction method for laser pulse width | |
| CN102288306B (en) | Method for simultaneously measuring output single-pulse energy and waveforms of lasers | |
| CN107561497B (en) | FSK and multiple non-linear frequency modulation signal identification and parameter estimation method | |
| WO2016004687A1 (en) | Method for distinguishing initial time point of ultra-high-frequency partial discharge signal | |
| CN109270346A (en) | Polarization state waveform signal processing method at a kind of OPGW lightning stroke | |
| CN111046327A (en) | Prony analysis method suitable for low-frequency oscillation and subsynchronous oscillation identification | |
| CN105223481A (en) | Based on the partial-discharge ultrahigh-frequency signal initial time defining method of difference energy function | |
| CN105447243A (en) | Weak signal detection method based on adaptive fractional order stochastic resonance system | |
| CN108089097B (en) | Intelligent online distribution network ground fault positioning method | |
| Xu et al. | Match-extracting chirplet transform with application to bearing fault diagnosis | |
| CN113687433A (en) | Bi-LSTM-based magnetotelluric signal denoising method and system | |
| CN115797772A (en) | Classified early warning method and system for icing thickness of overhead transmission line | |
| Parshin et al. | Usage of non-Gaussian statistics for RF signals detection by complex energy and fractal detector | |
| CN104734793B (en) | The energy detection method of wireless cooperation frequency spectrum perception based on p powers | |
| CN111638402B (en) | Method and system for measuring frequency of microwave signal in microwave electric field | |
| CN104458586A (en) | Method for screening ring-down signals in optical feedback cavity ring-down technology | |
| CN111814578B (en) | Method for extracting frequency of ultralow frequency Doppler signal | |
| CN113411122B (en) | Solar blind ultraviolet light communication self-adaptive signal detection method based on deep 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 |