CN114417742A - Laser atmospheric flicker index prediction method and system - Google Patents
Laser atmospheric flicker index prediction method and system Download PDFInfo
- Publication number
- CN114417742A CN114417742A CN202210335492.4A CN202210335492A CN114417742A CN 114417742 A CN114417742 A CN 114417742A CN 202210335492 A CN202210335492 A CN 202210335492A CN 114417742 A CN114417742 A CN 114417742A
- Authority
- CN
- China
- Prior art keywords
- flicker index
- neural network
- atmospheric
- flicker
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 54
- 238000013528 artificial neural network Methods 0.000 claims abstract description 51
- 238000003062 neural network model Methods 0.000 claims abstract description 40
- 238000004088 simulation Methods 0.000 claims abstract description 37
- 238000013527 convolutional neural network Methods 0.000 claims description 38
- 238000012360 testing method Methods 0.000 claims description 20
- 238000009826 distribution Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 11
- 230000007547 defect Effects 0.000 abstract description 5
- 230000005540 biological transmission Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 5
- 238000013507 mapping Methods 0.000 description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000001228 spectrum Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000000523 sample Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
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
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Photometry And Measurement Of Optical Pulse Characteristics (AREA)
Abstract
The invention discloses a laser atmospheric flicker index prediction method and a system, comprising the following steps: constructing an atmospheric turbulence simulation model; recording a light intensity pattern of the reference light beam after passing through the atmospheric turbulence simulation model, and calculating a flicker index of the reference light beam after passing through the atmospheric turbulence simulation model; the light intensity pattern and the flicker index are used as training data and are sent to a neural network for training to obtain a trained neural network model; and (4) sending the light intensity pattern of the unknown light beam into a neural network model, and predicting the flicker index of the unknown light beam. The method overcomes the defects of complex calculation, long time consumption, difficulty in eliminating the influence of external noise factors, poor robustness and the like in the prior art, and realizes the respective prediction of each atmospheric flicker index in the atmospheric models with different turbulence intensities through the neural network. The prediction is accurate, the calculation is simple, and the influence of external noise factors is avoided.
Description
Technical Field
The invention relates to the field of laser engineering and optical communication, in particular to a laser atmospheric flicker index prediction method and a laser atmospheric flicker index prediction system.
Background
To study the effect of atmospheric turbulence on laser propagation, it is important to calculate the laser atmospheric flicker index. At present, the laser atmospheric scintillation index is mainly calculated by the following method:
(1) a random phase screen is used to simulate atmospheric turbulence. The turbulence influence in the continuous space is equivalent to the superposition of a plurality of equally spaced turbulence phase screens, and then the actual transmission of the light beam in the atmosphere is simulated through a turbulence random phase screen model and an angular spectrum transmission formula. Calculating the scintillation index of the beam according to the following formula:
in the formula, the tip brackets indicate the ensemble average,Iindicating the intensity of the light at the probe end.
According to the method, numerical simulation can be carried out on the light intensity and the flicker index of the laser beam after the laser beam passes through the atmospheric turbulence.
(2) And estimating the atmospheric flicker index by adopting a projection optical method. The method is based on multiplicative modulation hypothesis of light source intensity fluctuation and atmospheric flicker, light intensity flicker on two different receiving apertures is measured at the same time through projection optics, and a measurement model is solved by combining aperture smoothing factors under the weak fluctuation condition, so that atmospheric flicker indexes are respectively estimated. This method needs to eliminate the effect of background light intensity and detector random noise.
The defects of the method for calculating the laser atmospheric flicker index are as follows: 1. the simulation calculation is complex and the consumed time is long; 2. the influence of external factors (noise and the like) is difficult to completely eliminate in experimental measurement; 3. these methods are not robust, and the calculation results are only applicable to the current turbulence situation, and need to be recalculated in other turbulence situations.
Disclosure of Invention
The invention aims to solve the technical problems that in the prior art, the atmospheric flicker index is complex to calculate, long in time consumption, difficult to eliminate the influence of external noise factors, poor in robustness and the like, and aims to provide a laser atmospheric flicker index prediction method and a laser atmospheric flicker index prediction system.
The invention is realized by the following technical scheme:
a laser atmospheric flicker index prediction method comprises the following steps: step S1: constructing an atmospheric turbulence simulation model; recording a light intensity pattern of the reference light beam after passing through the atmospheric turbulence simulation model, and calculating a flicker index of the reference light beam after passing through the atmospheric turbulence simulation model; step S2: the light intensity pattern and the flicker index are used as training data and are sent to a neural network for training to obtain a trained neural network model; and predicting the flicker index of the unknown light beam through the trained neural network model.
The method takes the existing data (including the light intensity pattern and the flicker index corresponding to the light intensity pattern) generated by the reference light beam as training data to train the neural network so as to form a mature atmospheric flicker index neural network model, and then sends the light intensity pattern of the unknown light beam into the atmospheric flicker index neural network model, so that the flicker index of the unknown light beam can be quickly obtained, and the quick prediction of the flicker index is realized. The invention adopts the neural network to learn and predict, realizes the quick and simple calculation of the atmospheric flicker index, avoids the influence of external noise factors and has good robustness.
The core concept of the invention is that a neural network model is adopted, and the intelligent prediction of the flicker index is realized. Wherein the atmospheric turbulence simulation model can be constructed according to the data of the prior art.
Further, the neural network comprises a convolutional neural network, and the neural network model comprises a convolutional neural network model; the step S2 specifically includes: taking the light intensity pattern of the reference beam as the input of the convolutional neural network, taking the flicker index of the reference beam as the output of the convolutional neural network, and training the convolutional neural network; and sending the light intensity pattern of the unknown light beam into the convolutional neural network model to obtain the flicker index of the unknown light beam.
The known training data is adopted to train the convolutional neural network, and the mapping relation between the light intensity pattern and the flicker index is represented through the convolutional neural network model, so that the prediction of the flicker index of the unknown light beam is realized.
Further, the neural network further comprises a time-series neural network, and the neural network model further comprises a time-series neural network model; the step S1 specifically includes: recording the flicker indexes of the reference light beams after passing through the atmospheric turbulence simulation model in continuous time to form a flicker index sequence in continuous time; the step S2 specifically includes: sending the flicker index sequence in the continuous time into the time sequence neural network, and training the time sequence neural network; and predicting the flicker index sequence of the future time period of the light beam through the time sequence neural network model.
The known training data is adopted to train the time sequence neural network, and the mapping relation between the flicker index sequences of different time periods in front and back is represented by the time sequence neural network model, so that the technical effect of predicting the flicker index sequence of the next time period by the flicker index sequence of the previous time period is realized.
Further, dividing the flicker index sequence in the continuous time into a flicker index sequence in a previous time interval and a flicker index sequence in a later time interval according to time; taking the flicker index sequence of the previous time interval as a training set, and training and correcting the time sequence neural network according to a sliding window strategy to obtain a time sequence neural network model; and taking the flicker index sequence of the later period as a test set, and verifying the time sequence neural network model.
Further, the length ratio of the flicker index sequence of the previous period to the flicker index sequence of the next period is: 8.5: 1.5.
further, in step S1, constructing an atmospheric turbulence simulation model, including: and simulating the atmospheric turbulence by adopting a random phase screen, and constructing the position distribution of the random phase screen according to the atmospheric model.
Further, the turbulence intensity of the atmospheric turbulence is used
Is shown to be in the range of 10-18-10-14(ii) a Will range from 10-18-10-14Is divided into five turbulence intensities by order of magnitude. The atmospheric flicker index of the reference beam at each turbulence intensity can be trained by using a neural network model. The atmospheric model simulates five different orders of magnitude turbulence intensity.
Further, in step S1, recording a light intensity pattern of the reference light beam after passing through the atmospheric turbulence simulation model, and calculating a flicker index of the reference light beam after passing through the atmospheric turbulence simulation model, specifically: and recording a light intensity pattern of the reference light beam passing through the atmospheric turbulence simulation model under each turbulence intensity, and calculating a flicker index of the reference light beam passing through the atmospheric turbulence simulation model under each turbulence intensity.
Further, the reference beam is a simulated gaussian beam. And the simulated Gaussian beam is adopted, so that the calculation process of the flicker index in the training data is simplified, and the influence of external noise factors is avoided.
In a second implementation manner of the present invention, a laser atmospheric flicker index prediction system includes a processor and a machine-readable storage medium, where the machine-readable storage medium stores machine-executable instructions, and the machine-executable instructions are loaded and executed by the processor to implement the laser atmospheric flicker index prediction method.
Compared with the prior art, the invention has the following advantages and beneficial effects:
the defects of complex calculation, long time consumption, difficulty in eliminating the influence of external noise factors, poor robustness and the like in the prior art are overcome, and the atmospheric flicker indexes in the atmospheric models with different turbulence intensities are respectively predicted through the convolutional neural network. The prediction is accurate, the calculation is simple, and the influence of external noise factors is avoided; meanwhile, the atmospheric flicker index in the future period is predicted through a time sequence neural network. The invention provides a novel prediction method applied to laser atmospheric flicker indexes.
Drawings
In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and that for those skilled in the art, other related drawings can be obtained from these drawings without inventive effort. In the drawings:
FIG. 1 is a flow chart of a method for predicting laser atmospheric flicker index using a convolutional neural network;
FIG. 2 is a flow chart of a method for predicting laser atmospheric flicker index using a temporal neural network;
FIG. 3 is a schematic diagram of the spot intensity pattern of a simulated laser after five intensity turbulences (where the numerical values represent the turbulence intensity);
FIG. 4 is a schematic diagram of laser atmospheric flicker index results obtained by using a convolutional neural network for atmospheric turbulence to be slowly changed (turbulence intensity is within the same order of magnitude), including flicker index predicted values, true values, and maximum and minimum value intervals;
FIG. 5 shows the atmospheric turbulence to be varied (turbulence intensity in different orders of magnitude, 10)-18-10-16) The schematic diagram of the prediction result of the laser atmospheric flicker index obtained by using the convolutional neural network comprises a flicker index prediction value,The real value and the maximum and minimum value intervals;
FIG. 6 is a graph of a prediction result of a laser atmospheric flicker index obtained by using a time sequence neural network for atmospheric turbulence to be slowly changed (turbulence intensity is within the same order of magnitude), including a flicker index prediction value and a true value;
FIG. 7 is a graph of atmospheric turbulence to be varied (turbulence intensity on different orders of magnitude, 10)-18-10-16) A laser atmospheric flicker index sequence prediction result curve graph obtained by using a time sequence neural network comprises a flicker index prediction value and a true value;
FIG. 8 is a schematic diagram of a flicker index sequence trained by a time-series neural network.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.
Example 1
The embodiment 1 is a laser atmospheric flicker index prediction method, which includes the following steps:
step S1: constructing an atmospheric turbulence simulation model, recording a light intensity pattern of a reference light beam after the reference light beam passes through the atmospheric turbulence simulation model, and calculating a flicker index of the reference light beam after the reference light beam passes through the atmospheric turbulence simulation model;
step S2: the light intensity pattern and the flicker index are used as training data and are sent to a neural network for training to obtain a trained neural network model; and predicting the flicker index of the unknown light beam by using the trained neural network model.
In this embodiment 1, existing data (including a light intensity pattern and a flicker index corresponding to the light intensity pattern) generated by a reference light beam is used as training data to train a neural network to form a mature atmospheric flicker index neural network model, and then the light intensity pattern of an unknown light beam is sent into the atmospheric flicker index neural network model, so that a flicker index of the unknown light beam can be quickly obtained, and thus, the flicker index can be quickly predicted. In this embodiment 1, a neural network is used for learning and predicting, so that the atmospheric flicker index is quickly and simply calculated, the influence of external noise factors is avoided, and robustness is good. The core concept of this embodiment 1 is that a neural network model is used to realize intelligent prediction of flicker index. Wherein the atmospheric turbulence simulation model can be constructed according to the data of the prior art.
In one possible embodiment, the neural network comprises a convolutional neural network, and the neural network model comprises a convolutional neural network model; step S2 specifically includes: taking the light intensity pattern of the reference beam as the input of a convolutional neural network, taking the flicker index of the reference beam as the output of the convolutional neural network, and training the convolutional neural network; and sending the light intensity pattern of the unknown light beam into a convolutional neural network model to obtain the flicker index of the unknown light beam. The convolutional neural network is trained through known training data, and the mapping relation between the light intensity pattern and the flicker index is represented through a convolutional neural network model, so that the prediction of the flicker index of the unknown light beam is realized.
In one possible embodiment, the neural network further comprises a time-series neural network, and the neural network model further comprises a time-series neural network model; step S1 specifically includes: recording the flicker indexes of the reference light beams after passing through the atmospheric turbulence simulation model in continuous time to form a flicker index sequence in continuous time; step S2 specifically includes: sending the flicker index sequence in continuous time into a time sequence neural network, and training the time sequence neural network; and predicting the flicker index sequence of the future time period of the light beam through a time sequence neural network model.
Dividing the flicker index sequence in the continuous time into a flicker index sequence in a previous time interval and a flicker index sequence in a next time interval according to time; taking the flicker index sequence of the previous time interval as a training set, and training and correcting the time sequence neural network according to a sliding window strategy to obtain a time sequence neural network model; and then, taking the flicker index sequence of the later period as a test set, and verifying the time sequence neural network model.
Each sliding window is a flicker index window sequence with the same length, the last flicker index of the sliding window sequence is used as the output of the time sequence neural network, and other flicker index sequences except the last flicker index in the window sequence are used as the input of the time sequence neural network, so that the time sequence neural network is trained and corrected. That is, in the training of the time-series neural network, only the last flicker index in a sliding window can be used as an output, and other flicker index sequences in the sliding window are used as inputs.
In one possible embodiment, the length ratio in time of the preceding period flicker index sequence (i.e., training set) to the following period flicker index sequence (i.e., test set) is 8.5: 1.5.
for example: a total of 1000 scintillation indices, before training 8.5: the 1.5 ratio is divided into a training sequence and a test sequence, i.e. the first 850 flicker indexes are used for training, and the last 150 flicker indexes are used for testing, i.e. the length ratio of the flicker index sequence of the previous time interval and the flicker index sequence of the next time interval is referred to in this embodiment. During training, using the 850 flicker index data, as shown in fig. 8, 10 length data (the first solid frame in the figure) are taken each time, the first 9 data are used as input, and the 10 th data (gray dots) are used as output. Then, the solid line frame is moved one step later (changed into a short horizontal line frame in the figure), the frame is selected to have the next 10 lengths, the first 9 are used as input again, and the 10 th is used as output. Then the short transverse wire frame is pushed one step backwards, and so on. After training is finished, the first 9 of 150 flicker index test data are used as time sequence neural network input, and the time sequence neural network model can sequentially predict the values and trends of the later 141 flicker index data, namely the time sequence neural network learns the mapping relation from 850 flicker index data and can predict the values of a period of continuous time in the future. The reliability of the time series neural network model was verified with 150 flicker index test data.
In this example, the training set sliding window holds 10 flicker indices, the first 9 as inputs and the last 1 as outputs. Under different application scenes, the sliding window holds N flicker indexes, the Nth flicker index is used as the output of neural network training, and a sequence formed by the 1 st to the N-1 st flicker indexes is used as the input of time sequence neural network training.
Therefore, even if each secondary sequence has 100 flicker indexes, 99 are used as inputs and 100 is used as an output, and then the next secondary sequence is 2 to 101 flicker indexes, 2 to 100 input and 101 output. One flicker index value at a time.
In this embodiment, the length ratio of the preceding period flicker index sequence to the succeeding period flicker index sequence in terms of time is preferably 8.5:1.5, generally not more than 8: 2.
and predicting the flicker index sequence of the light beam in the future time period by utilizing the trained and corrected time sequence neural network model. The known training data is adopted to train the time sequence convolution neural network, and the mapping relation between the flicker indexes of different continuous time periods before and after the time sequence convolution neural network model is represented, so that the technical effect of predicting the flicker index sequence of the next time period by the flicker index sequence of the previous time period is realized.
In one possible embodiment, in step S1, an atmospheric turbulence simulation model is constructed, including: and simulating the atmospheric turbulence by adopting a random phase screen, and constructing the position distribution of the random phase screen according to the atmospheric model. The atmosphere model here is an atmosphere model that is already available in the prior art.
For intensity of turbulence of atmospheric turbulence
Is shown to be in the range of 10-18-10-14(ii) a Will range from 10-18-10-14Is divided into five turbulence intensities by order of magnitude. The atmospheric flicker index of the reference beam at each turbulence intensity can be trained by using a neural network model. The atmospheric model simulates five different orders of magnitude turbulence intensity.
In one possible embodiment, step S1 includes: and recording a light intensity pattern of the reference light beam passing through the atmospheric turbulence simulation model under each turbulence intensity, and calculating a flicker index of the reference light beam passing through the atmospheric turbulence simulation model under each turbulence intensity. And the prediction of the atmospheric flicker index under different turbulence intensities is realized.
In this embodiment 1, the reference beam is a simulated gaussian beam. And the simulated Gaussian beam is adopted, so that the calculation process of the flicker index in the training data is simplified, and the influence of external noise factors is avoided. In addition, scintillation index
Is calculated as follows:
wherein, the sharp brackets indicate the ensemble average,Iindicating the intensity of light received at the probe end.
Example 2
The embodiment 2 is a method for predicting a laser atmospheric flicker index based on the embodiment 1, and the method comprises the following steps:
step A: in this embodiment 2, a random phase screen is used to simulate an atmospheric turbulence, and the transmission of a light beam in the atmosphere is equivalent to "free space transmission + phase screen", and the distribution of the phase screens is constructed according to an atmospheric model.
1. Random phase screen simulation process:
(1) firstly, random complex numbers which are obeyed standard normal distribution are put into an NxN matrix
Performing the following steps;
(2) which is then filtered with a random phase spectrum.
Namely:
. Wherein:
which represents a spectrum of random phases,
representing an NxN matrix
，
(3) and performing inverse Fourier transform on the obtained result, and taking the real part as the random phase loaded by the random phase screen.
2. Simulation of the transmission process of the light beam in the atmosphere:
according to the simulation process of the above phase screen, the change of the light field caused by the p-th phase screen can be described as:
(1) Computing
Two-dimensional Fourier transform in the x-y plane
；
(2) Multiplying the obtained result by
To obtain
；
(3) Obtaining the light field of
；
(4) The light intensity can be obtained by utilizing the obtained light field, and the light intensity pattern of the light beam after the light beam passes through the turbulent flow is obtained.
And B: recording a light intensity pattern of the reference beam after the reference beam passes through turbulence and calculating a flicker index; simulating a light intensity pattern of the Gaussian beam after turbulence and recording; the scintillation index of the beam at each turbulence intensity was calculated. A schematic diagram of the spot intensity pattern of simulated Gaussian light after being subjected to five intensity turbulences is shown in FIG. 3, in which the value is 10-18-10-14The range represents the turbulence intensity.
Intensity of turbulence
Is 10-18-10-14Range, divided into five orders of magnitude turbulence intensity. To pairSimulating a light intensity pattern of the Gaussian beam passing through the random phase screen according to the step A and recording the light intensity pattern at each turbulence intensity; and calculating the flicker index of the light beam under each turbulence intensity according to the light intensity pattern. The scintillation index is calculated according to the following formula:
wherein, the sharp brackets indicate the ensemble average,Iindicating the total intensity of the light intensity pattern,Nrepresenting the number of intensity patterns used to calculate the ensemble average.
And C: establishing a neural network model; example 2 uses a convolutional neural network to correct turbulence intensity (10) over a range-18-10-14Any subrange or all) of the light beam scintillation indices; and also learning the light beam flicker index passing through the turbulent flow in the past period of time by adopting a time sequence neural network, and predicting the light beam flicker index passing through the turbulent flow in the future period of time. Specifically, the following flicker index sequence of length 286 is predicted from the previous flicker index sequence (length 1684), and the ratio of the sequence length of the flicker index to the predicted sequence length is 8.5:1.5, but preferably not more than 8: 2.
the two methods of the present embodiment predict flicker index. First, a convolutional neural network is used to couple the passing turbulence (intensity 10)-18-10-14Either sub-range) is predicted. And B, forming a data set by using the light intensity pattern and the flicker index data pair obtained in the step B, wherein the light intensity pattern is input by a network, the flicker index is a training label, and the training convolution network can predict the flicker index according to the light intensity pattern. A flow chart of a method for predicting laser atmospheric flicker index by using a convolutional neural network is shown in fig. 1. Secondly, a time sequence neural network is adopted to learn the scintillation index sequence trend collected and calculated in the continuous time of the light beam, and the light beam scintillation index sequence in a period of time in the future is predicted (the ratio of the length of the learned sequence to the length of the predicted sequence is 8.5:1.5, and the ratio can be adjusted, but preferably does not exceed 8: 2). Laser atmospheric flicker index using time sequential neural networkA flow chart of the prediction method is shown in fig. 2.
Step D: network training; training a convolutional neural network by adopting a light intensity pattern and a flicker index recorded in a simulation mode, and setting the learning rate to 10 by adopting an Adam optimizer-4Calculating the first order moment estimate and the exponential decay factor of the second order moment estimate to be set to 0.5 and 0.999, and setting the loss function to be mae; training a time sequence neural network by using a flicker index recorded by simulation, and adopting an Adam optimizer.
And training a convolutional neural network and a time sequence neural network by using the light intensity pattern and the flicker index obtained by the simulation experiment. When the convolutional neural network is trained, the objective function is as follows:
wherein,
a convolutional neural network is represented that is,
in order to initialize the parameters for the network,
obtaining optimal network parameters for training;
representing a light intensity map of the light source,
is the true flicker index. The convolution network adopts an Adam gradient descent optimization algorithm and aims to reduce the error between the predicted flicker index and the real flicker index. During training, the learning rate is set to 10-4The loss function is set to mean absolute error:
。
and when the time sequence neural network is trained, training by using the flicker index sequence recorded by simulation. The scintillation index sequence was as follows 8.5: a scale of 1.5 is divided into a training set and a test set. And (3) taking out a sequence window with the length of 10 every time from the training sequence, taking the first 9 values of the sequence as network input, taking the 10 th value as a training label, moving the window according to 1 step length every time, and inputting data to train the network. When the time sequence network is trained, an Adam gradient descent optimization algorithm is also adopted, and a loss function is set as an average absolute error.
Step E: testing a network; adopting a light intensity pattern which is not seen by the network during training as an input test convolutional neural network; test data with a training window length is input into a time sequence neural network, and the network predicts and outputs a future specified number of light beam flicker indexes under the current turbulence intensity trend.
And inputting the trained network by using the test set data to obtain the predicted flicker index. For the convolutional neural network, the light intensity pattern of the test set is used as input to predict the flicker index; for the time sequence neural network, the test sequence is input into the time sequence neural network, and the network predicts and outputs the light beam flicker index of a future period under the current turbulence intensity trend.
The embodiment 2 overcomes the defects of complex calculation, long time consumption, difficulty in eliminating the influence of external noise factors, poor robustness and the like in the prior art, and provides a novel prediction method applied to laser atmospheric flicker indexes.
Example 3
This example 3 is based on example 2,
1. and carrying out numerical simulation on the light intensity and the flicker index of the reference beam after the reference beam passes through the atmospheric turbulence by using MATLAB. The effect of atmospheric turbulence on the beam is simulated using a random phase screen. The method specifically comprises the following steps: the light beam is transmitted through a free space at a distance, then a phase plate is added, and the distribution of the phase plate is constructed according to an atmospheric model; then, the transmission is carried out in a free space with a certain distance, and a phase plate is added; a transmission of a similar structure is performed for a certain number of cycles.
2. Simulating light intensity diagram of Gaussian beam after passing through different intensity atmospheric turbulences, wherein the intensity of turbulence is used
Is shown to be in the range of 10-18-10-14. Wherein, the turbulence intensity in each order of magnitude is equally divided according to 2000, and the simulated light beam passes through 2000 intensity turbulences in one order of magnitude. For better calculation of the flicker index, the average value of the light intensity of 500 light beam transmissions is recorded as the ensemble average value under the same conditions (constant turbulence intensity), i.e. as in equation (1)<I>Substituting the formula to calculate the flicker index. Finally, 2000 flicker indexes are calculated by simulating turbulent intensity light beam transmission of one order of magnitude. Meanwhile, one light intensity pattern in 500 transmissions is selected and put into a data set, and finally turbulent intensity light beam transmission of one order of magnitude is simulated to obtain 2000 light intensity patterns. The intensity map data set and flicker index data thus collected will be used to train the convolutional neural network and the time series neural network.
3. Preparation work before training the convolutional neural network: and forming a data pair by the acquired light intensity graph and the flicker index. And calculating the mean value, the maximum value and the minimum value of the flicker indexes of the 15 light intensity maps before and after each light intensity map to serve as the flicker index of the current light intensity map and the maximum value and the minimum value thereof, wherein the most value can be used for evaluating the network prediction result. Training the convolutional neural network, extracting the light intensity map and the corresponding flicker index from the data set at equal intervals (such as every 150 sheets) to form a test set, and forming a training set by the rest data. For a time-series neural network, the flicker index data set (i.e., flicker index sequence) is adjusted according to a ratio of 8.5: a scale of 1.5 is divided into a training set and a test set.
4. And respectively establishing a convolutional neural network and a time sequence neural network. For a convolutional neural network, inputting a light intensity map with the size of 64 multiplied by 64, adopting Adam by an optimization algorithm, training iteration times of 100 times, storing trained model parameters, and inputting the light intensity map in a test set into the trained model to obtain a predicted flicker index; for the time sequence neural network, a flicker index sequence with a specified length is input (the length is determined according to data quantity, for example, 2000 flicker indexes, the length of an input window can be 10), Adam is adopted in an optimization algorithm, the number of training iterations is 2, trained model parameters are stored, then a test sequence is input into a trained model, and the trend and the numerical value of the flicker index sequence with the length 15% compared with the training sequence are predicted.
5. And testing the trained model by using the test set. As shown in FIG. 4, the result was 10-17A flicker index predicted by the convolutional neural network within a turbulence intensity range; the network test result curve chart comprises a flicker index predicted value, a true value, a maximum value interval and a minimum value interval; the error rate table of the corresponding network test result is as follows:
| actual flicker index | Prediction of flicker index | Relative error rate |
| 1.72×10-5 | 1.54×10-5 | -10.47% |
| 2.64×10-5 | 2.93×10-5 | 10.98% |
| 3.51×10-5 | 3.38×10-5 | -3.70% |
| 4.40×10-5 | 4.88×10-5 | 10.91% |
| 5.44×10-5 | 6.66×10-5 | 22.43% |
| 6.16×10-5 | 6.59×10-5 | 6.98% |
| 7.38×10-5 | 8.28×10-5 | 12.20% |
| 8.31×10-5 | 7.32×10-5 | -11.91% |
| 8.99×10-5 | 7.96×10-5 | -11.46% |
| 9.86×10-5 | 9.22×10-5 | -6.49% |
It can be seen that by using the trained convolutional network, a good flicker index prediction result can be obtained for the slowly-varying atmospheric turbulence. Wherein the error rate of 90% is below 30%.
As shown in fig. 5, the convolutional network has a higher prediction accuracy for a large flicker index, and has a lower prediction accuracy for a small flicker index, i.e., a bias occurs. For large atmospheric turbulences to be varied (turbulence intensity in different orders of magnitude, 10)-18-10-16) Make itAs shown in fig. 5, a schematic diagram of a prediction result of a laser atmospheric flicker index obtained by using a convolutional neural network shows that an error table of a corresponding network test result is as follows:
| actual flicker index | Prediction of flicker index | Relative error rate |
| 1.72×10-5 | 1.74×10-5 | 1.41% |
| 2.64×10-5 | 3.64×10-5 | 38.08% |
| 3.51×10-5 | 3.63×10-5 | 3.45% |
| 4.40×10-5 | 4.98×10-5 | 13.09% |
| 5.44×10-5 | 5.56×10-5 | 2.15% |
| 6.16×10-5 | 5.19×10-5 | -15.77% |
| 7.38×10-5 | 6.12×10-5 | -17.03% |
| 8.31×10-5 | 7.52×10-5 | -9.51% |
| 8.99×10-5 | 8.01×10-5 | -10.90% |
| 9.86×10-5 | 8.32×10-5 | -15.64% |
| 1.70×10-4 | 1.48×10-4 | -12.94% |
| 2.57×10-4 | 2.18×10-4 | -15.18% |
| 3.59×10-4 | 2.40×10-4 | -33.15% |
| 4.54×10-4 | 3.04×10-4 | -33.04% |
| 5.37×10-4 | 4.61×10-4 | -14.15% |
| 6.38×10-4 | 6.22×10-4 | -2.51% |
| 7.39×10-4 | 6.45×10-4 | -12.72% |
| 8.35×10-4 | 7.48×10-4 | -10.42% |
| 8.96×10-4 | 7.61×10-4 | -15.07% |
| 1.01×10-3 | 8.93×10-4 | -11.58% |
| 1.70×10-3 | 2.10×10-3 | 23.53% |
| 2.56×10-3 | 3.22×10-3 | 25.78% |
| 3.59×10-3 | 3.19×10-3 | -11.14% |
| 4.54×10-3 | 4.20×10-3 | -7.49% |
| 5.37×10-3 | 5.74×10-3 | 6.89% |
| 6.38×10-3 | 7.37×10-3 | 15.52% |
| 7.36×10-3 | 7.25×10-3 | -1.49% |
| 8.32×10-3 | 8.10×10-3 | -2.64% |
| 8.92×10-3 | 8.15×10-3 | -8.63% |
| 1.01×10-2 | 1.04×10-2 | 2.97% |
As shown in fig. 6-7, the timing network compensates the deviation defect of the convolutional neural network, and more accurately predicts the variation trend of the flicker index along with the variation of the turbulence. FIG. 6 is a graph of a prediction result of a laser atmospheric flicker index obtained by using a time sequence neural network for atmospheric turbulence to be slowly changed (turbulence intensity is within the same order of magnitude), including a flicker index prediction value and a true value; the error rate table of partial network test results is as follows:
| true flicker index | Prediction of flicker index | Relative error rate |
| 1.10×10-2 | 1.28×10-2 | 16.36% |
| 1.10×10-2 | 1.21×10-2 | 10.00% |
| 1.96×10-2 | 1.70×10-2 | -13.27% |
| 1.32×10-2 | 1.35×10-2 | 2.27% |
| 1.46×10-2 | 9.96×10-3 | -31.78% |
| 1.74×10-2 | 1.41×10-2 | -18.97% |
| 2.70×10-2 | 2.11×10-2 | -21.85% |
| 3.00×10-2 | 2.51×10-2 | -16.33% |
| 3.08×10-2 | 2.91×10-2 | -5.52% |
| 3.75×10-2 | 3.42×10-2 | -8.80% |
| 4.01×10-2 | 3.88×10-2 | -3.24% |
| 3.93×10-2 | 3.78×10-2 | -3.82% |
| 3.96×10-2 | 4.10×10-2 | 3.54% |
| 3.44×10-2 | 3.62×10-2 | 5.23% |
| 3.15×10-2 | 3.05×10-2 | -3.17% |
FIG. 7 is a graph of atmospheric turbulence to be varied (turbulence intensity on different orders of magnitude, 10)-18-10-16) A laser atmospheric flicker index sequence prediction result curve graph obtained by using a time sequence neural network comprises a flicker index prediction value and a true value; the error rate table of partial network test results is as follows:
| true flicker index | Prediction of flicker index | Relative error rate table |
| 1.39×10-2 | 1.58×10-2 | 13.67% |
| 6.65×10-3 | 7.98×10-3 | 20.00% |
| 1.30×10-2 | 1.55×10-2 | 19.38% |
| 7.43×10-3 | 5.25×10-3 | -29.34% |
| 1.03×10-2 | 7.20×10-3 | -30.10% |
| 1.05×10-2 | 1.29×10-2 | 22.86% |
| 7.09×10-3 | 6.87×10-3 | -3.10% |
| 1.22×10-2 | 1.50×10-2 | 23.28% |
| 1.51×10-2 | 1.29×10-2 | -14.57% |
| 1.12×10-2 | 1.29×10-2 | 15.18% |
| 1.48×10-2 | 1.42×10-2 | -4.05% |
| 1.60×10-2 | 1.87×10-2 | 16.63% |
| 1.21×10-2 | 1.40×10-2 | 15.70% |
| 8.45×10-3 | 1.04×10-2 | 23.08% |
| 9.23×10-3 | 1.20×10-2 | 30.01% |
6. The software and hardware equipment adopted for training is as follows: 4 GeForce2080 display card, Ubuntu18.04.3 operating system, Python3.6 programming language, TensorFlow1.8.0, keras2.1.6 deep learning framework, Vscode compiling environment.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A laser atmospheric flicker index prediction method is characterized by comprising the following steps:
step S1: constructing an atmospheric turbulence simulation model; recording a light intensity pattern of the reference light beam after passing through the atmospheric turbulence simulation model, and calculating a flicker index of the reference light beam after passing through the atmospheric turbulence simulation model;
step S2: the light intensity pattern and the flicker index are used as training data and are sent to a neural network for training to obtain a trained neural network model;
and predicting the flicker index of the unknown light beam through the trained neural network model.
2. The laser atmospheric flicker index prediction method of claim 1, wherein the neural network comprises a convolutional neural network, and the neural network model comprises a convolutional neural network model;
the step S2 specifically includes: taking the light intensity pattern of the reference beam as the input of the convolutional neural network, taking the flicker index of the reference beam as the output of the convolutional neural network, and training the convolutional neural network;
and sending the light intensity pattern of the unknown light beam into the convolutional neural network model to obtain the flicker index of the unknown light beam.
3. The laser atmospheric flicker index prediction method of claim 1, wherein the neural network further comprises a time-series neural network, and the neural network model further comprises a time-series neural network model;
the step S1 specifically includes: recording the flicker indexes of the reference light beams after passing through the atmospheric turbulence simulation model in continuous time to form a flicker index sequence in continuous time;
the step S2 specifically includes: sending the flicker index sequence in the continuous time into the time sequence neural network, and training the time sequence neural network; and predicting the flicker index sequence of the future time period of the light beam through the time sequence neural network model.
4. The laser atmospheric flicker index prediction method according to claim 3, wherein the flicker index sequence in the continuous time is divided into a flicker index sequence in a previous period and a flicker index sequence in a subsequent period in time;
taking the flicker index sequence of the previous time interval as a training set, and training and correcting the time sequence neural network according to a sliding window strategy to obtain a time sequence neural network model;
and taking the flicker index sequence of the later period as a test set, and verifying the time sequence neural network model.
5. The laser atmospheric flicker index prediction method of claim 4, wherein the length ratio of the flicker index sequence of the previous period to the flicker index sequence of the subsequent period is as follows: 8.5: 1.5.
6. the laser atmospheric flicker index prediction method of claim 1, wherein in the step S1, the constructing an atmospheric turbulence simulation model comprises: and simulating the atmospheric turbulence by adopting a random phase screen, and constructing the position distribution of the random phase screen according to the atmospheric model.
7. The laser atmospheric flicker index prediction method of claim 6, wherein the turbulence intensity of the atmospheric turbulence is used
Is shown to be in the range of 10-18-10-14(ii) a Will range from 10-18-10-14Is divided into five turbulence intensities by order of magnitude.
8. The laser atmospheric flicker index prediction method according to claim 7, wherein in the step S1, the light intensity pattern of the reference beam after passing through the atmospheric turbulence simulation model is recorded, and the flicker index of the reference beam after passing through the atmospheric turbulence simulation model is calculated, specifically:
and recording a light intensity pattern of the reference light beam passing through the atmospheric turbulence simulation model under each turbulence intensity, and calculating a flicker index of the reference light beam passing through the atmospheric turbulence simulation model under each turbulence intensity.
9. The laser atmospheric flicker index prediction method of any one of claims 1 to 8, wherein the reference beam is a simulated gaussian beam.
10. A laser atmospheric flicker index prediction system, comprising a processor and a machine-readable storage medium having machine-executable instructions stored therein, the machine-executable instructions being loaded and executed by the processor to implement the laser atmospheric flicker index prediction method of any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210335492.4A CN114417742B (en) | 2022-04-01 | 2022-04-01 | Laser atmospheric flicker index prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210335492.4A CN114417742B (en) | 2022-04-01 | 2022-04-01 | Laser atmospheric flicker index prediction method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114417742A true CN114417742A (en) | 2022-04-29 |
| CN114417742B CN114417742B (en) | 2022-06-10 |
Family
ID=81263074
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210335492.4A Active CN114417742B (en) | 2022-04-01 | 2022-04-01 | Laser atmospheric flicker index prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114417742B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115374712A (en) * | 2022-10-24 | 2022-11-22 | 中国航天三江集团有限公司 | Method and device for calibrating optical transmission simulation parameters under influence of laser internal channel thermal effect |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160026865A1 (en) * | 2014-07-24 | 2016-01-28 | EXELlS, INC. | Vision-based system for dynamic weather detection |
| CN109100759A (en) * | 2018-08-01 | 2018-12-28 | 东南大学 | A kind of ionosphere Amplitude scintillation detection method based on machine learning |
| CN111639767A (en) * | 2020-05-27 | 2020-09-08 | 西安电子科技大学 | Optical orbit angular momentum machine learning identification method based on turbulence effect |
| AU2020102396A4 (en) * | 2020-09-23 | 2020-11-05 | Changchun University Of Science And Technology | Machine learning-based method for analyzing characteristics of laser beam propagation through turbulent atmosphere |
| US20200387579A1 (en) * | 2019-06-10 | 2020-12-10 | General Electric Company | Deep learning surrogate for turbulent flow |
| CN112083444A (en) * | 2020-08-19 | 2020-12-15 | 中国民用航空总局第二研究所 | Low-latitude airport ionosphere short-time prediction method considering plasma bubbles |
| CN113222250A (en) * | 2021-05-13 | 2021-08-06 | 中国工程物理研究院激光聚变研究中心 | High-power laser device output waveform prediction method based on convolutional neural network |
| WO2022002917A1 (en) * | 2020-06-29 | 2022-01-06 | Unistellar | Method and system for estimating a characteristic atmospheric turbulence parameter |
-
2022
- 2022-04-01 CN CN202210335492.4A patent/CN114417742B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160026865A1 (en) * | 2014-07-24 | 2016-01-28 | EXELlS, INC. | Vision-based system for dynamic weather detection |
| CN109100759A (en) * | 2018-08-01 | 2018-12-28 | 东南大学 | A kind of ionosphere Amplitude scintillation detection method based on machine learning |
| US20200387579A1 (en) * | 2019-06-10 | 2020-12-10 | General Electric Company | Deep learning surrogate for turbulent flow |
| CN111639767A (en) * | 2020-05-27 | 2020-09-08 | 西安电子科技大学 | Optical orbit angular momentum machine learning identification method based on turbulence effect |
| WO2022002917A1 (en) * | 2020-06-29 | 2022-01-06 | Unistellar | Method and system for estimating a characteristic atmospheric turbulence parameter |
| CN112083444A (en) * | 2020-08-19 | 2020-12-15 | 中国民用航空总局第二研究所 | Low-latitude airport ionosphere short-time prediction method considering plasma bubbles |
| AU2020102396A4 (en) * | 2020-09-23 | 2020-11-05 | Changchun University Of Science And Technology | Machine learning-based method for analyzing characteristics of laser beam propagation through turbulent atmosphere |
| CN113222250A (en) * | 2021-05-13 | 2021-08-06 | 中国工程物理研究院激光聚变研究中心 | High-power laser device output waveform prediction method based on convolutional neural network |
Non-Patent Citations (5)
| Title |
|---|
| ARTEM M. VORONTSOV等: ""Atmospheric Turbulence Study with Deep Machine Learning of Intensity Scintillation Patterns"", 《APPLIED SCIENCES》, vol. 10, no. 22, 17 November 2020 (2020-11-17), pages 1 - 26 * |
| 张骏昕等: ""云南高美古湍流大气中激光短曝光光斑统计特性研究"", 《中国激光》, vol. 48, no. 1, 31 January 2021 (2021-01-31), pages 1 - 10 * |
| 徐福隆: ""基于深度学习的高纬度电离层闪烁预测"", 《中国优秀硕士学位论文全文数据（电子期刊）》, no. 1, 15 January 2021 (2021-01-15), pages 008 - 210 * |
| 王佳斌等: "激光在湍流大气中传输的闪烁系数及其测量", 《强激光与粒子束》 * |
| 王佳斌等: "激光在湍流大气中传输的闪烁系数及其测量", 《强激光与粒子束》, vol. 23, no. 04, 30 April 2011 (2011-04-30), pages 915 - 918 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115374712A (en) * | 2022-10-24 | 2022-11-22 | 中国航天三江集团有限公司 | Method and device for calibrating optical transmission simulation parameters under influence of laser internal channel thermal effect |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114417742B (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107977710B (en) | Electricity consumption abnormal data detection method and device | |
| CN114417742B (en) | Laser atmospheric flicker index prediction method and system | |
| CN101739819A (en) | Method and device for predicting traffic flow | |
| AU2020102396A4 (en) | Machine learning-based method for analyzing characteristics of laser beam propagation through turbulent atmosphere | |
| CN112464567B (en) | Intelligent data assimilation method based on variational and assimilative framework | |
| Song et al. | Fractional Brownian motion: Difference iterative forecasting models | |
| CN111898734B (en) | NMR relaxation time inversion method based on MLP | |
| CN112116002A (en) | Determination method, verification method and device of detection model | |
| CN112214880B (en) | Multi-stage maintainability assessment method based on virtual-real fusion | |
| CN106972949A (en) | A kind of fractional order network system situation method of estimation based on adaptive equalization technology | |
| CN117391463A (en) | River pollution tracing method | |
| WO2020105468A1 (en) | Information processing device, information processing system, information processing method, and non-transitory computer-readable medium having program stored therein | |
| CN103279030B (en) | Dynamic soft measuring modeling method and device based on Bayesian frame | |
| KR20200028801A (en) | Learning method and learning device for variational interference using neural network and test method and test device for variational interference using the same | |
| CN115598714B (en) | Time-space coupling neural network-based ground penetrating radar electromagnetic wave impedance inversion method | |
| CN110991660A (en) | Situation analysis method, system and storage medium of LSSVM-ARIMA model based on locust optimization | |
| CN116299702A (en) | CNN-based frequency domain low-frequency expansion multi-scale full waveform inversion method | |
| CN112329535B (en) | CNN-based quick identification method for low-frequency oscillation modal characteristics of power system | |
| CN110995203B (en) | Nonlinear observability degree analysis method based on condition number fusion | |
| Robbiati et al. | A quantum analytical Adam descent through parameter shift rule using Qibo | |
| CN115047408A (en) | Underwater multi-sound-source positioning method based on single-layer large convolution kernel neural network | |
| CN118378037B (en) | Bridge acceleration data denoising method based on long-short-term memory depth neural network | |
| Lippe et al. | Modeling accurate long rollouts with temporal neural PDE solvers | |
| CN113779873B (en) | Construction method and application of TDECQ measurement model | |
| KR102464449B1 (en) | Apparatus and method for Laplace Fourier domain full waveform inversion using multiple attenuation and multiple offsets |
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 |