CN117631029A - Rayleigh surface wave dispersion curve inversion method based on joint algorithm - Google Patents
Rayleigh surface wave dispersion curve inversion method based on joint algorithm Download PDFInfo
- Publication number
- CN117631029A CN117631029A CN202410107332.3A CN202410107332A CN117631029A CN 117631029 A CN117631029 A CN 117631029A CN 202410107332 A CN202410107332 A CN 202410107332A CN 117631029 A CN117631029 A CN 117631029A
- Authority
- CN
- China
- Prior art keywords
- neural network
- layer
- inversion
- model
- algorithm
- 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.)
- Pending
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 58
- 239000006185 dispersion Substances 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013528 artificial neural network Methods 0.000 claims abstract description 67
- 238000005457 optimization Methods 0.000 claims abstract description 50
- 238000012549 training Methods 0.000 claims abstract description 36
- 230000006870 function Effects 0.000 claims description 14
- 239000002245 particle Substances 0.000 claims description 13
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 3
- 239000010410 layer Substances 0.000 description 40
- 230000007547 defect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000011229 interlayer Substances 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
Abstract
The invention discloses a Rayleigh surface wave dispersion curve inversion method and system based on a joint algorithm, comprising the steps of constructing a stratum model, performing forward modeling on the stratum model, and generating a training sample; constructing a neural network, inputting the generated training sample into the constructed neural network, and training the neural network; inputting the observation data into the trained neural network to obtain a prediction result; and taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a swarm intelligent optimization algorithm. The method and the device continue to optimize on the result predicted by the neural network, and ensure the accuracy of the inversion result; the initial solution provided by the neural network has rationality, can indicate the direction for the subsequent iteration of the group intelligent optimization algorithm, and simultaneously greatly reduces the iteration times.
Description
Technical Field
The invention relates to the technical field of exploration data processing based on artificial intelligence, in particular to a Rayleigh surface wave dispersion curve inversion method based on a joint algorithm.
Background
The Rayleigh surface wave is formed by mutual interference of P wave and SV wave and propagates along a free interface, and has the characteristics of strong energy, slow attenuation, high signal to noise ratio and the like. The exploration method based on the Rayleigh waves utilizes the dispersion characteristics of the Rayleigh waves to estimate the transverse wave speed, and is widely applied to various fields such as shallow stratum engineering geological investigation, low-speed zone identification, cavity investigation, large-structure investigation and the like.
The Rayleigh surface wave exploration comprises the stages of data acquisition, processing, interpretation and the like. The Rayleigh surface wave dispersion curve inversion is a key ring of a processing stage, and whether the exploration result is accurate or not is directly determined. In the Rayleigh surface wave inversion method widely applied at present, the gradient algorithm mainly comprises a least square method, an OCCAM method and the like, and the global optimization algorithm mainly comprises a particle swarm algorithm, a genetic algorithm, a simulated annealing method, a neural network algorithm and the like. However, in these methods, the gradient algorithm depends on the selection of the initial solution, and the swarm intelligent optimization algorithm such as the particle swarm algorithm is slow to iterate and has poor precision in the complex inversion process. In recent years, because of the development of artificial intelligence, a neural network algorithm becomes an important inversion method, but the neural network algorithm also has the defects of dependence on the selection of network training samples, incapability of continuing optimization and the like. Therefore, the invention provides a Rayleigh surface wave dispersion curve inversion method based on the combination of a neural network and a swarm intelligent optimization algorithm. The method overcomes the defect that the neural network cannot continue to be optimized, and overcomes the defects that the group intelligent optimization algorithm is slow in iteration and low in convergence accuracy when the initial solution is not provided.
Disclosure of Invention
Therefore, the invention aims to provide a Rayleigh surface wave dispersion curve inversion method and system based on a joint algorithm. And (3) continuing to carry out optimization on the result predicted by the neural network, and ensuring the accuracy of the inversion result.
Therefore, the invention provides a Rayleigh surface wave dispersion curve inversion method based on a neural network and intelligent optimization combined algorithm, which comprises the following steps:
s1, constructing a stratum model, and performing forward modeling on the stratum model to generate a training sample;
s2, constructing a neural network, inputting the generated training sample into the constructed neural network, and training the neural network;
s3, inputting the observation data into the trained neural network to obtain a prediction result;
and S4, taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a swarm intelligent optimization algorithm.
Further preferably, in S1, the building a stratum model and performing forward modeling on the stratum model to generate a training sample includes the following steps:
s101, dividing the whole stratum model into a plurality of thin layers;
s102, assigning the transverse wave speed of each thin layer in a mode of gradually increasing the transverse wave speed layer by layer;
s103, setting random probability and selecting a disturbance layer to carry out random disturbance to form a model sample set;
s104, carrying out forward modeling on each layer of model in the model sample set to obtain a dispersion curve corresponding to each layer of model.
Further preferably, in S102, the method further includes:
for any firstLayer transverse wave velocityDegree->According to the following formula (1), the relation between the transverse wave speed and the longitudinal wave speed of each layer is established, and according to the formula (2), the relation between the transverse wave speed and the density is established:
(1)
(2)
wherein,for the k-th layer longitudinal wave velocity, +.>Is the k layer density; />Is the i-th layer thickness.
Further preferably, in S103, the model sample set includes an incremental speed model based, a high speed sandwich model, and a low speed sandwich model.
Further preferably, in S2, the neural network is a BP neural network, and the neural network includes, when constructed:
setting a three-layer network structure comprising an input layer, a plurality of hidden layers and an output layer;
taking a dispersion curve as input and a stratum transverse wave speed as output;
and setting a transmission function between the input layer and the hidden layer as a log function, and setting a transmission function between the hidden layer and the output layer as a pureline function.
Further preferably, in S4, the swarm intelligent optimization algorithm employs a particle swarm algorithm,
and taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a particle swarm algorithm.
Setting inversion parameters, taking a prediction result as a historical global optimal solution of the particle swarm, and guiding the searching direction of the particle swarm; the inversion parameters include: population, maximum iteration number, preset precision, inertia weight, learning factor and iteration stop criterion.
The invention also provides a Rayleigh surface wave dispersion curve inversion system based on the neural network and the intelligent optimization joint algorithm, which comprises the following steps: the system comprises a stratum model construction module, a neural network prediction module and an optimization algorithm inversion module;
the stratum model construction module is used for constructing a stratum model, performing forward modeling on the stratum model and generating a training sample;
the neural network prediction module is used for inputting the generated training samples into the constructed neural network and training the neural network; inputting the observation data into the trained neural network to obtain a prediction result;
the optimization algorithm inversion module; and taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a swarm intelligent optimization algorithm.
The present invention also provides an electronic device including:
a memory storing computer program instructions;
and the processor, when the computer program instructions are executed by the processor, realizes the steps of the Rayleigh wave dispersion curve inversion method based on the neural network and intelligent optimization combined algorithm.
The invention also provides a computer readable storage medium for storing instructions that, when executed on a computer, cause the computer to perform the steps of the rayleigh wave dispersion curve inversion method based on the neural network and the intelligent optimization joint algorithm.
Compared with the prior art, the Rayleigh surface wave dispersion curve inversion method and system based on the joint algorithm continue to optimize on the result predicted by the neural network, and ensure the accuracy of the inversion result; since the swarm intelligence optimization algorithm often requires more iterations without an initial solution, and is prone to falling into local optima. The initial solution provided by the neural network has rationality, can indicate the direction for the subsequent iteration of the group intelligent optimization algorithm, and simultaneously greatly reduces the iteration times. When the neural network prediction result is the optimal solution, the group intelligent optimization algorithm does not need to continue iterative optimization, so that the calculation efficiency is greatly improved.
Drawings
FIG. 1 is a schematic structural diagram of a Rayleigh surface wave dispersion curve inversion method based on a neural network and an intelligent optimization combined algorithm;
FIG. 2 is a diagram of a BP neural network training process of the present invention;
FIG. 3 is a graph of a dispersion curve fit of the inversion result of the incremental velocity model of the present invention;
FIG. 4 is a graph of a high-speed interlayer model inversion result dispersion curve fit;
FIG. 5 is a graph of a low-speed sandwich model inversion result dispersion curve fit;
FIG. 6 is a graph of the inversion results of an incremental velocity model provided by the present invention;
FIG. 7 is a diagram of the inversion results of the high-speed sandwich model provided by the invention;
FIG. 8 is a graph of inversion results of a low-speed sandwich model provided by the invention.
Detailed Description
The invention is described in further detail below with reference to the drawings and the detailed description.
As shown in fig. 1, the rayleigh wave dispersion curve inversion method based on the neural network and intelligent optimization combined algorithm provided by the embodiment of the invention comprises the following steps:
s1, constructing a stratum model, and performing forward modeling on the stratum model to generate a training sample;
in S1, the method specifically includes the following steps:
s101, dividing the whole stratum model into a plurality of thin layers;
s102, assigning the transverse wave speed of each thin layer in a mode of gradually increasing the transverse wave speed layer by layer;
s103, setting random probability and selecting a disturbance layer to carry out random disturbance to form a model sample set; forming a model sample set which is mainly an incremental speed model and simultaneously contains a small amount of high-speed interlayer models and low-speed interlayer models;
further include, for any of the firstLayer transverse wave speed +.>According to the following formula (1) and formula (2), the relation between the transverse wave speed and the longitudinal wave speed of each layer and the relation between the transverse wave speed and the density are established:
(1)
(2)
s105, forward modeling is conducted on each layer of model in the model sample set, and a dispersion curve corresponding to each layer of model is obtained. Wherein each layer of model and the corresponding dispersion curve together form a set of complete samples.
S2, constructing a neural network, inputting the generated training sample into the constructed neural network, and training the neural network; further, in S2, the neural network includes, but is not limited to, one of a BP neural network, a convolutional neural network, a deep belief network. Before training, firstly determining the number of layers of a neural network, the number of neurons of each layer and an activation function adopted among the layers, determining a network learning rate, determining network input as observation data, namely a dispersion curve obtained by forward modeling, determining network output as model parameters, namely the transverse wave speed of each stratum, determining the proportion of training samples, verification samples and test samples in the samples, and setting a network training stopping criterion at least to (1) achieve the maximum training times, (2) ensure that the error between the predicted output and the theoretical output is smaller than a training target, and (3) ensure that one of weight and bias update does not occur in continuous prescribed times of training.
In the method, a BP neural network is selected as an inversion network, a three-layer network structure is set, a dispersion curve is used as input, the stratum shear wave speed is used as output, the transfer function between an input layer and a hidden layer and between the hidden layer and the hidden layer is a log sig function, the transfer function between the hidden layer and an output layer is a pureline function, a network training function is a quantized conjugate gradient algorithm, and the proportion of a training sample, a verification sample and a test sample is 8:1:1, training stop conditions include: (1) the maximum training times are reached, (2) the error between the predicted output and the theoretical output is smaller than the training target, and (3) the weight and bias update no longer occurs in 20 continuous training. After the network parameter setting is completed, a sample set is introduced, and training is started. The network training results are shown in fig. 2.
S3, inputting the observation data into the trained neural network to obtain a prediction result (inversion dispersion curve), and introducing the obtained prediction result as an initial solution into a swarm intelligent optimization algorithm to perform optimization inversion.
And introducing the observation data into the BP neural network, and outputting a prediction result through the trained mapping relation.
And introducing the predicted result as an initial solution into a swarm intelligent optimization algorithm to develop further optimization inversion. The swarm intelligent optimization algorithm adopts a particle swarm algorithm (PSO), and before inversion is carried out, inversion parameters are firstly set, including: population, maximum iteration number, preset precision, inertia weight, learning factor and iteration stop criterion. When PSO is introduced into the BP neural network prediction result serving as an initial solution, the BP neural network prediction result serving as a historical global optimal solution of the particle swarm is used for guiding the searching direction of the particle swarm.
The present invention uses three different formation models to conduct experiments, namely a velocity increment model shown in table 1, a high-velocity sandwich model shown in table 2, and a low-velocity sandwich model shown in table 3. The inversion results given in tables 1-3 and figures 3-8 show that the PSO inversion result is closer to the theoretical model result, so that the combination of the neural network and the swarm intelligent optimization algorithm can have higher reliability than the independent use of one method.
TABLE 1 incremental speed model and inversion results thereof
TABLE 2 high speed Sandwich model and inversion results thereof
TABLE 3 Low speed Sandwich model and inversion results thereof
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (9)
1. The Rayleigh surface wave dispersion curve inversion method based on the neural network and intelligent optimization joint algorithm is characterized by comprising the following steps of:
s1, constructing a stratum model, and performing forward modeling on the stratum model to generate a training sample;
s2, constructing a neural network, inputting the generated training sample into the constructed neural network, and training the neural network;
s3, inputting the observation data into the trained neural network to obtain a prediction result;
and S4, taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a swarm intelligent optimization algorithm.
2. The method for inverting the rayleigh wave dispersion curve based on the neural network and intelligent optimization combined algorithm according to claim 1, wherein in S1, the formation model is constructed, forward modeling is performed on the formation model, and a training sample is generated, comprising the following steps:
s101, dividing the whole stratum model into a plurality of thin layers;
s102, assigning the transverse wave speed of each thin layer in a mode of gradually increasing the transverse wave speed layer by layer;
s103, setting random probability and selecting a disturbance layer to carry out random disturbance to form a model sample set;
s104, carrying out forward modeling on each layer of model in the model sample set to obtain a dispersion curve corresponding to each layer of model.
3. The method for inverting the rayleigh wave dispersion curve based on the neural network and intelligent optimization joint algorithm according to claim 1, wherein in S102, further comprising:
for any firstLayer transverse wave speed +.>According to the following formula (1), the relation between the transverse wave speed and the longitudinal wave speed of each layer is established, and according to the formula (2), the relation between the transverse wave speed and the density is established:
(1)
(2)
wherein,for the k-th layer longitudinal wave velocity, +.>Is the k layer density; />Is the i-th layer thickness.
4. The method for the inversion of a rayleigh wave dispersion curve based on a neural network and intelligent optimization joint algorithm according to claim 2, wherein in S103, the model sample set includes an incremental velocity model high-speed sandwich model and a low-speed sandwich model.
5. The method for inverting the rayleigh wave dispersion curve based on the combined algorithm of the neural network and the intelligent optimization according to claim 1, wherein in S2, the neural network adopts a BP neural network, and the neural network comprises:
setting a three-layer network structure comprising an input layer, a plurality of hidden layers and an output layer;
taking a dispersion curve as input and a stratum transverse wave speed as output;
and setting a transmission function between the input layer and the hidden layer as a log function, and setting a transmission function between the hidden layer and the output layer as a pureline function.
6. The method for inverting the rayleigh wave dispersion curve based on the neural network and intelligent optimization combined algorithm according to claim 1, which is characterized by comprising the following steps of: in S4, the intelligent optimization algorithm of the group adopts a particle swarm algorithm,
taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a particle swarm algorithm;
setting inversion parameters, taking a prediction result as a historical global optimal solution of the particle swarm, and guiding the searching direction of the particle swarm; the inversion parameters include: population, maximum iteration number, preset precision, inertia weight, learning factor and iteration stop criterion.
7. A Rayleigh surface wave dispersion curve inversion system based on a neural network and intelligent optimization joint algorithm is characterized by comprising: the system comprises a stratum model construction module, a neural network prediction module and an optimization algorithm inversion module;
the stratum model construction module is used for constructing a stratum model, performing forward modeling on the stratum model and generating a training sample;
the neural network prediction module is used for inputting the generated training samples into the constructed neural network and training the neural network; inputting the observation data into the trained neural network to obtain a prediction result;
the optimization algorithm inversion module; and taking the obtained prediction result as an initial solution, and carrying out optimization inversion by using a swarm intelligent optimization algorithm.
8. An electronic device, comprising:
a memory storing computer program instructions;
a processor, which when executed by the processor, implements the steps of the rayleigh wave dispersion curve inversion method based on a neural network and intelligent optimization joint algorithm as claimed in any one of claims 1 to 6.
9. A computer readable storage medium storing instructions which, when executed on a computer, cause the computer to perform the steps of the rayleigh wave dispersion curve inversion method based on a neural network and intelligent optimization joint algorithm according to any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410107332.3A CN117631029A (en) | 2024-01-26 | 2024-01-26 | Rayleigh surface wave dispersion curve inversion method based on joint algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410107332.3A CN117631029A (en) | 2024-01-26 | 2024-01-26 | Rayleigh surface wave dispersion curve inversion method based on joint algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117631029A true CN117631029A (en) | 2024-03-01 |
Family
ID=90030780
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410107332.3A Pending CN117631029A (en) | 2024-01-26 | 2024-01-26 | Rayleigh surface wave dispersion curve inversion method based on joint algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117631029A (en) |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110879412A (en) * | 2019-10-31 | 2020-03-13 | 南方科技大学 | Underground transverse wave velocity inversion method, device, computing equipment and storage medium |
CN113327322A (en) * | 2021-05-26 | 2021-08-31 | 国勘数字地球(北京)科技有限公司 | Inversion method and device of surface wave frequency dispersion curve and computer readable storage medium |
CN113642232A (en) * | 2021-07-22 | 2021-11-12 | 南方科技大学 | Surface wave intelligent inversion exploration method, storage medium and terminal equipment |
CN114779324A (en) * | 2022-03-11 | 2022-07-22 | 西南交通大学 | Rayleigh wave frequency dispersion curve inversion method based on deep learning and application |
WO2022227206A1 (en) * | 2021-04-26 | 2022-11-03 | 中国科学院地质与地球物理研究所 | Fully convolutional neural network-based magnetotelluric inversion method |
CN115453623A (en) * | 2022-09-20 | 2022-12-09 | 电子科技大学长三角研究院(湖州) | UNet-based receiving function and surface wave frequency dispersion joint inversion method |
US20230084403A1 (en) * | 2020-01-14 | 2023-03-16 | Schlumberger Technology Corporation | Machine learning driven dispersion curve picking |
CN116299665A (en) * | 2022-09-07 | 2023-06-23 | 西南石油大学 | LSTM surface wave inversion method, device and medium |
CN116663428A (en) * | 2023-07-24 | 2023-08-29 | 中国地震局地球物理研究所 | Neural network-based face wave imaging joint inversion method |
-
2024
- 2024-01-26 CN CN202410107332.3A patent/CN117631029A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110879412A (en) * | 2019-10-31 | 2020-03-13 | 南方科技大学 | Underground transverse wave velocity inversion method, device, computing equipment and storage medium |
US20230084403A1 (en) * | 2020-01-14 | 2023-03-16 | Schlumberger Technology Corporation | Machine learning driven dispersion curve picking |
WO2022227206A1 (en) * | 2021-04-26 | 2022-11-03 | 中国科学院地质与地球物理研究所 | Fully convolutional neural network-based magnetotelluric inversion method |
CN113327322A (en) * | 2021-05-26 | 2021-08-31 | 国勘数字地球(北京)科技有限公司 | Inversion method and device of surface wave frequency dispersion curve and computer readable storage medium |
CN113642232A (en) * | 2021-07-22 | 2021-11-12 | 南方科技大学 | Surface wave intelligent inversion exploration method, storage medium and terminal equipment |
CN114779324A (en) * | 2022-03-11 | 2022-07-22 | 西南交通大学 | Rayleigh wave frequency dispersion curve inversion method based on deep learning and application |
CN116299665A (en) * | 2022-09-07 | 2023-06-23 | 西南石油大学 | LSTM surface wave inversion method, device and medium |
CN115453623A (en) * | 2022-09-20 | 2022-12-09 | 电子科技大学长三角研究院(湖州) | UNet-based receiving function and surface wave frequency dispersion joint inversion method |
CN116663428A (en) * | 2023-07-24 | 2023-08-29 | 中国地震局地球物理研究所 | Neural network-based face wave imaging joint inversion method |
Non-Patent Citations (1)
Title |
---|
张志厚等: "瑞雷波频散曲线的深度学习反演方法", 《地球物理学报》, vol. 65, no. 6, 30 June 2022 (2022-06-30), pages 2244 - 2259 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109799533A (en) | A kind of method for predicting reservoir based on bidirectional circulating neural network | |
CN110618453B (en) | Wave impedance inversion method based on improved damping least square method | |
CN107100611B (en) | Double-layer intelligent drilling speed modeling method in geological exploration drilling process | |
CN114330644B (en) | Neural network model compression method based on structure search and channel pruning | |
CN114723095A (en) | Missing well logging curve prediction method and device | |
CN113610945A (en) | Ground stress curve prediction method based on hybrid neural network | |
CN110991284A (en) | Optical remote sensing image statement description generation method based on scene pre-classification | |
CN113722980A (en) | Ocean wave height prediction method, system, computer equipment, storage medium and terminal | |
CN115841076A (en) | Shallow sea layered seabed ground sound parameter inversion method based on BP neural network model | |
CN111208563A (en) | Non-split complete matching layer absorption boundary method | |
CN113486591B (en) | Gravity multi-parameter data density weighted inversion method for convolutional neural network result | |
CN116663428B (en) | Neural network-based face wave imaging joint inversion method | |
CN114637881A (en) | Image retrieval method based on multi-agent metric learning | |
CN117631029A (en) | Rayleigh surface wave dispersion curve inversion method based on joint algorithm | |
CN109143346B (en) | Pre-stack hybrid nonlinear inversion method and computer-readable storage medium | |
Yan et al. | Elastic parameter inversion problem based on brain storm optimization algorithm | |
CN110441815B (en) | Simulated annealing Rayleigh wave inversion method based on differential evolution and block coordinate descent | |
CN107798380A (en) | Ephemeris computational methods and computing system on detector device | |
CN115983105B (en) | Occam inversion Lagrange multiplier optimization method based on deep learning weighted decision | |
Syaharuddin et al. | Best Architecture Recommendations of ANN Backpropagation Based on Combination of Learning Rate, Momentum, and Number of Hidden Layers | |
CN113219534B (en) | Pre-stack depth migration speed quality control method and device, medium and electronic equipment | |
CN111749675A (en) | Stratum drillability prediction method and system based on cascade model algorithm | |
CN117892096B (en) | Small sample ocean sound velocity profile forecasting method based on transfer learning | |
Peng et al. | BDS-3 Broadcast Ephemeris Orbit Correction Model Based on Improved PSO Combined with BP Neural Network | |
CN112819140B (en) | OpenCL-based FPGA one-dimensional signal recognition neural network acceleration method |
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 |