CN111523258A - MS-Net network-based microseism effective signal first arrival pickup method and system - Google Patents

Abstract

The invention relates to the technical field of microseismic data processing, in particular to a method and system for picking up the first arrivals of microseismic effective signals based on an MS-Net network. The method includes generating an original data set; calibrating the data set; inputting the data set into the constructed MS-Net network for training, and obtaining optimal network model parameters, specifically including performing calibration on the part of the samples for calibration. Supervised training, the other part of the samples that have not been calibrated are subjected to unsupervised training; the probability distribution of the data set is calculated point by point; the system includes a data set production module, a data set training module and an output module; the embodiment of the present invention uses MS The ‑Net network combines the semi-supervised method to construct the total loss by weighting the unsupervised loss and the supervised loss. By minimizing the total loss, the network model parameters are optimized, and finally the accurate prediction and identification of the first arrival point of the effective signal are realized; the training set is reduced. Label the number of labels to improve training set quality and detection accuracy.

Description

Method and system for first-arrival picking of effective microseismic signals based on MS-Net network

technical field

The invention relates to the technical field of microseismic data processing, in particular to a method and system for picking up the first arrivals of microseismic effective signals based on an MS-Net network.

Background technique

The microseismic detection method plays an important role in engineering construction and geological disaster prevention and control. At the same time, the microseismic signal has the characteristics of weak signal energy and easy interference from background noise, resulting in the inability to accurately pick up the first arrival of the microseismic signal, resulting in microseismic events. Therefore, the effective signal detection method of microseismic is one of the key points in the field of microseismic data processing.

The traditional signal detection technology includes spectrum analysis of the signal by fast Fourier transform, time-frequency conversion by wavelet, curvelet and shearlet transform to achieve the purpose of removing noise and retaining effective signals. However, if the traditional method is directly applied to microseismic data, it is often unable to obtain satisfactory results, which will directly affect the quality and accuracy of microseismic monitoring. Signal monitoring based on deep learning has gradually attracted widespread attention in recent years. The main reason is that it has the characteristics of many parameters and large capacity, which makes its network have strong processing ability for massive data; MS-Net's new network model It is composed of Denseblock (Gao Huang, Zhuang Liu, Laurens van derMaaten, Kilian Q.Weinberger.2017) blocks added to the UNet++ network to deepen the network structure. Through the skip layer and pruning structure in the UNet++ network, the main and subtle signals are extracted. At the same time, it avoids the problems of feature stacking and over-fitting. By adding Denseblock blocks, it makes up for the problem that the recognition of deep features is not obvious due to the small number of UNet++ network layers, so that the construction of deep and shallow features can be accurately obtained. The MS-Net network improves the refined extraction of signal features to a certain extent.

The disadvantage of the prior art is that it needs to manually label the label data set and input it into the network for reinforcement training and learning, and the quality of the training set is not high, time-consuming and low in accuracy.

SUMMARY OF THE INVENTION

In order to overcome the deficiencies of the prior art, the embodiments of the present invention provide a method and system for first-arrival pickup of effective microseismic signals based on an MS-Net network, which reduces the number of labels in the training set and improves the quality and detection accuracy of the training set.

On the one hand, an embodiment of the present invention provides an MS-Net network-based microseismic effective signal first-arrival picking method, comprising the following steps:

S1, generate the original data set; specifically, the use of finite difference forward modeling to generate a large number of analog signals with a main frequency range of 20-1000 Hz under different models and the actual data together constitute the original data set;

S2, data set calibration; specifically, it includes first-arrival picking for a part of the samples of the original data set, selecting the first-arrival and non-first-arrival signal waveforms of each signal sampling point and performing calibration respectively, and the other part of the samples is not calibrated;

S3, input the data set into the constructed MS-Net network for training, and obtain optimal network model parameters; specifically, supervised training is performed on the part of the samples that have been calibrated, and the other part of the sample that has not been calibrated sample for unsupervised training;

S4, calculating the probability distribution of the data set point by point; specifically, using the softmax function to output the probability point by point, obtaining the binary classification probability of all points, and selecting the probability peak of the first arrival category as the first arrival point.

On the other hand, an embodiment of the present invention provides an MS-Net network-based microseismic effective signal first-arrival pickup system, including:

The data set production module generates the original data set; it specifically includes the use of finite difference forward modeling to generate a large number of analog signals with a main frequency range of 20 to 1000 Hz and the actual data to form the original data set; data set calibration; specifically including the original data Collect a part of the samples for first-arrival pick-up, select the signal waveforms at the first-arrival and non-first-arrival points of each signal sampling point and perform calibration respectively, and the other part of the samples are not calibrated;

Data set training module, input the data set into the constructed MS-Net network for training, and obtain the optimal network model parameters; specifically, supervised training is performed on the part of the samples that have been calibrated, and all the samples that have not been calibrated have been trained. The other part of the sample is used for unsupervised training;

The output module calculates the probability distribution of the data set point by point; specifically includes using the softmax function to output the probability point by point, obtaining the binary classification probability of all points, and selecting the probability peak of the first arrival category as the first arrival point.

Embodiments of the present invention provide a method and system for picking up the first arrivals of microseismic effective signals based on MS-Net network. By combining the MS-Net network with the semi-supervised method Temporal Ensembling, the unsupervised loss and the supervised loss are weighted and summed to construct a total loss. , by minimizing the total loss, optimizing the network model parameters, and finally realizing the accurate prediction and identification of the first arrival point of the effective signal; reducing the number of labels in the training set, improving the quality of the training set and detection accuracy.

Description of drawings

In order to illustrate the technical solutions of the present invention more clearly, the following briefly introduces the accompanying drawings used in the technical description of the present invention. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, which are of great significance to the art For those of ordinary skill, other drawings can also be obtained from these drawings without creative labor.

1 is a schematic flowchart of a method for picking up the first arrivals of microseismic effective signals based on an MS-Net network according to an embodiment of the present invention;

2 is a schematic diagram of a semi-supervised method combined with an MS-Net network training process according to an embodiment of the present invention;

Fig. 3 is the first-arrival position probability prediction curve diagram of the effective signal of the MS-Net network according to the embodiment of the present invention;

4 is a schematic structural diagram of an MS-Net network-based microseismic effective signal first-arrival pickup system according to an embodiment of the present invention;

Reference number:

Dataset production module-1Dataset training module-2Output module-3

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

1 is a schematic flowchart of a method for picking up microseismic effective signals for the first time based on an MS-Net network according to an embodiment of the present invention; as shown in FIG. 1 , the method includes the following steps:

S1, generate the original data set; specifically, the use of finite difference forward modeling to generate a large number of analog signals with a main frequency range of 20-1000 Hz under different models and the actual data together constitute the original data set;

S2, data set calibration; specifically, it includes first-arrival picking for a part of the samples of the original data set, selecting the first-arrival and non-first-arrival signal waveforms of each signal sampling point and performing calibration respectively, and the other part of the samples is not calibrated;

S3, input the data set into the constructed MS-Net network for training, and obtain optimal network model parameters; specifically, supervised training is performed on the part of the samples that have been calibrated, and the other part of the sample that has not been calibrated sample for unsupervised training;

S4, calculating the probability distribution of the data set point by point; specifically, using the softmax function to output the probability point by point, obtaining the binary classification probability of all points, and selecting the probability peak of the first arrival category as the first arrival point.

Specifically, FIG. 3 is a graph showing the probability prediction curve of the first-arrival position of the effective signal of the MS-Net network according to the embodiment of the present invention; as shown in FIG. 3 , a part of the samples of the original data set are first-arrived, and the first-arrival point of each signal sampling point is selected. The signal waveforms at the arrival and non-initial arrivals are calibrated separately, and the other part of the sample is not calibrated; at the same time, these two parts of the sample are input into the network training model of the MS-Net network combined with the semi-supervised method for training, and supervised accordingly. Training and unsupervised training to obtain the optimal network model parameters, that is, the probability prediction curve of the first arrival position of the effective signal in the data set; calculate the probability distribution of the data set point by point, use the softmax function to output the probability point by point, and obtain the second probability of each point. Classification probability, only the probability peak of the first arrival category is selected as the first arrival point, and its formula is:

Among them, q _i (x) represents the prediction probability distribution that x belongs to different categories, and x represents each point of the output f(x) of the last layer of the fully convolutional network; when the prediction probability distribution q _i (x) is 0, it means It is not the first arrival point, and the value of i is 2; when the probability value q _i (x) is 1, it represents the first arrival point, and the value of i is 1; k(x) represents the category, and k=1, 2 represents the respective Represents two types of first solstice and non-first solstice.

FIG. 2 is a schematic diagram of the training process of the semi-supervised method combined with the MS-Net network according to the embodiment of the present invention; as shown in FIG. 2 , in the step S3, the optimal network model parameters specifically include: combining an unsupervised loss function and a supervised loss The weighted summation of the functions constructs a total loss function, and obtains the minimized total loss function; the value of the minimized total loss function is less than 0.1.

Specifically, in the semi-supervised method combined with the MS-Net network training process, the total loss function is:

where C is the number of different classes and B is the mini-batch index set; the evaluation results of the two branches are divided into two different stages: first the training set is classified without updating the weights, and then the network is subjected to different evaluations under the same input Augmented and missing training of , using the just-obtained predictions as targets for the unsupervised loss component.

中(其中a是集合动量项)，

是训练过程中多轮xi的预测值集成的结果。

包含来自先前训练时期的网络集合输出的加权平均值，但最近的时期的权重大于远处的时期。为了产生训练以v为目标，我们需要通过除以因子(1-at)来校正V中的启动偏差，得到

在此过程中，我们可将第一个训练时期的无监督权重函数W(t)指定为零。After each training, the network output v _j is accumulated to the output by updating the extracted feature vector V _j ←aV _j +(1-a)v _j .

in (where a is the collective momentum term),

is the result of the ensemble of predictions from multiple rounds of xi during the training process.

Contains a weighted average of network ensemble outputs from previous training epochs, but more recent epochs have more weight than distant epochs. To generate training targeting v, we need to correct the startup bias in V by dividing by a factor (1-at), giving

During this process, we can specify zero for the unsupervised weight function W(t) for the first training epoch.

通过平方差函数形成无监督loss，当前输出V_j与标记样本通过交叉熵函数形成有监督loss，两次的loss值通过加权求和构成网络模型的总loss值，通过最小化总loss值，进而得到最优网络模型。The integration result of the current output V _j and the predicted values of multiple rounds of xi

The unsupervised loss is formed by the squared difference function, and the current output V _j and the marked sample form the supervised loss through the cross-entropy function. get the optimal network model.

The embodiment of the present invention provides a method for picking up the first arrivals of effective microseismic signals based on the MS-Net network. Through the MS-Net network combined with the semi-supervised method Temporal Ensembling, the unsupervised loss and the supervised loss are weighted and summed to construct the total loss, Minimize the total loss, optimize the network model parameters, and finally achieve accurate prediction and identification of the first arrival point of the effective signal; reduce the number of labels in the training set, and improve the quality of the training set and detection accuracy.

Based on the above embodiment, FIG. 4 is a schematic structural diagram of an MS-Net network-based microseismic effective signal first-arrival pickup system according to an embodiment of the present invention; as shown in FIG. 4 , it includes:

Data set production module 1, to generate the original data set; specifically, the use of finite difference forward modeling to generate a large number of analog signals with a main frequency range of 20 to 1000 Hz under different models and the actual data together form the original data set; data set calibration; A part of the samples in the data set is picked for the first arrival, and the signal waveforms at the first arrival and non-first arrival of each signal sampling point are selected and calibrated respectively, and the other part of the samples are not calibrated;

Data set training module 2, input the data set into the constructed MS-Net network for training, and obtain the optimal network model parameters; specifically, supervised training is performed on the part of the samples that are calibrated, and the samples that are not calibrated are The other part of the samples is subjected to unsupervised training;

The output module 3 calculates the probability distribution of the data set point by point; specifically includes using the softmax function to output the probability point by point, obtaining the binary classification probability of all points, and selecting the probability peak of the first arrival category as the first arrival point.

The embodiment of the present invention provides a microseismic effective signal first-arrival picking system based on MS-Net network to perform the above method. Through MS-Net network combined with semi-supervised method Temporal Ensembling, unsupervised loss and supervised loss are weighted and summed to construct a total loss, by minimizing the total loss, optimizing the network model parameters, and finally realizing the accurate prediction and identification of the effective signal initial arrival point; reducing the number of labels in the training set, improving the quality of the training set and detection accuracy.

The device embodiments described above are only illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed over multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

From the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on this understanding, the above-mentioned technical solutions can be embodied in the form of software products in essence or the parts that make contributions to the prior art, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic A disc, an optical disc, etc., includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be The technical solutions described in the foregoing embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (4)

1. A microseism effective signal first arrival picking method based on an MS-Net network is characterized by comprising the following steps:

s1, generating an original data set; specifically, under different models generated by forward finite difference, a large number of analog signals with a dominant frequency range of 20-1000 Hz and actual data form an original data set together;

s2, calibrating a data set; the method specifically comprises the steps of carrying out first arrival picking on a part of samples of an original data set, selecting signal waveforms of first arrival and non-first arrival positions of each signal sampling point, and respectively calibrating, wherein the other part of samples are not calibrated;

s3, inputting the data set into the constructed MS-Net network for training to obtain the optimal network model parameters; specifically, the method comprises the steps of carrying out supervised training on one part of samples subjected to calibration, and carrying out unsupervised training on the other part of samples not subjected to calibration;

s4, calculating the probability distribution of the data set point by point; the method specifically comprises the steps of utilizing a softmax function to output probability point by point to obtain two classification probabilities of all points, and selecting a probability peak value of a first-arrival category as a first-arrival point.

2. The method for picking up the first arrival of the microseism valid signal based on the MS-Net network of claim 1, wherein in the step S3, the optimal network model parameters specifically include: weighting and summing an unsupervised loss function and a supervised loss function to construct a total loss function, and obtaining a minimized total loss function; minimizing the total loss function value to less than 0.1.

3. A microseism effective signal first arrival picking system based on an MS-Net network is characterized by comprising:

the data set production module generates an original data set; specifically, under different models generated by forward finite difference, a large number of analog signals with a dominant frequency range of 20-1000 Hz and actual data form an original data set together; calibrating a data set; the method specifically comprises the steps of carrying out first arrival picking on a part of samples of an original data set, selecting signal waveforms of first arrival and non-first arrival positions of each signal sampling point, and respectively calibrating, wherein the other part of samples are not calibrated;

the data set training module is used for inputting the data set into the constructed MS-Net network for training to obtain the optimal network parameters; specifically, the method comprises the steps of carrying out supervised training on one part of samples subjected to calibration, and carrying out unsupervised training on the other part of samples not subjected to calibration;

the output module calculates the probability distribution of the data set point by point; the method specifically comprises the steps of utilizing a softmax function to output probability point by point to obtain two classification probabilities of all points, and selecting a probability peak value of a first-arrival category as a first-arrival point.

4. The MS-Net network based microseism valid signal first arrival pickup system according to claim 3, wherein the data set training module specifically comprises the following parameters: weighting and summing an unsupervised loss function and a supervised loss function to construct a total loss function, and obtaining a minimized total loss function; minimizing the total loss function value to less than 0.1.

Images