CN117590461A - Deposition type identification model establishment, deposition type identification method and related device - Google Patents

Abstract

The invention discloses a deposition type identification model establishment method, a deposition type identification method and a related device. The method for establishing the deposition type identification model comprises the steps of obtaining a sample set according to a plurality of selected logging curves and deposition type data of at least one well, wherein samples in the sample set comprise curve segments and deposition types of the selected logging curves in a certain depth range of the corresponding well; and setting a network structure and network parameters of the deep neural network according to a set rule, and training the set deep neural network by adopting a periodic attenuation type learning rate through a sample set to obtain a deposit type identification model for predicting target well deposit type data. The model established by the method can rapidly and reasonably complete the prediction of the target well deposition type data based on the well data; the periodic attenuation type learning rate is used for optimizing the loss function, so that the convergence speed of the loss function is improved, the loss function has stronger capability of escaping from saddle points and local minimum values, and further the prediction efficiency and the prediction precision are improved.

Description

Deposition type identification model establishment, deposition type identification method and related device

Technical Field

The invention relates to the technical field of geophysics, in particular to a sediment type identification model establishment method, a sediment type identification method and a related device.

Background

Sedimentary facies and lithology classification are important research contents in geophysical data interpretation, provide technical support for reservoir prediction and reef discovery, and are the precondition for subsequent fine data interpretation. For example, the method can provide phase constraint conditions for subsequent inversion work so as to improve inversion accuracy, or is used for well shock calibration and layer sequence grid establishment so as to provide support for earthquake low-frequency prediction. In the geophysical field, conventional depositional facies and lithology classification methods mostly use means of waveform analysis, seismic attribute calculation, manual interpretation and the like of seismic data, and the limitations of the methods are shown in the following aspects: (1) The processing and interpretation links are more, the flow is complex, and errors generated by each link can be transmitted and accumulated step by step, so that the final precision is affected; (2) Too much depends on the data quality, the expertise of interpreters and the subjective judgment of the research work area; (3) less time-efficient in the face of large data volumes; (4) knowledge accumulation, promotion and migration use presents difficulties.

Disclosure of Invention

The inventor finds that in recent years, with the promotion of calculation power, related algorithm research is rising and a large amount of data is accumulated, so that an intelligent method shows efficiency and accuracy advantages exceeding those of human beings in some application fields such as classification, regression, dimension reduction, clustering and the like. Taking deep learning as an example, more nonlinear layers are included, the characteristic extraction capability and the expression capability are strong, the learning rate is taken as an important super parameter, and a small constant or attenuation type array is usually manually given for optimizing the loss function. However, when the random gradient descent optimization method is used in deep learning, the two learning rates may be set so that the found local minimum value and the true global minimum value have a large difference, and even the saddle point of the loss function may be trapped to affect the prediction accuracy of the model. In addition, to ensure success of model training, the learning rate is generally given to be small, which makes training efficiency low especially in the later stages of training. However, if a large learning rate is set by a person, the learning efficiency may be improved in the early stage, but the loss function may not be converged and training may fail.

Meanwhile, the well data has the characteristics of high longitudinal resolution and coverage of various physical property information of the underground. Therefore, the sedimentary facies and lithology classification are carried out by utilizing a deep learning intelligent method based on well data so as to meet the production requirements that exploration targets are increasingly hidden and exploration precision requirements are increasingly improved.

In view of the foregoing, the present invention has been made to provide a deposition type identification model creation, deposition type identification method, and related apparatus that overcomes or at least partially solves the foregoing problems, by which the created model is capable of quickly and reasonably completing prediction of target well deposition type data based on well data; the periodic attenuation type learning rate is used for optimizing the loss function, so that the convergence speed of the loss function is improved, the loss function has stronger capability of escaping from saddle points and local minimum values, and further the prediction precision is improved.

In a first aspect, an embodiment of the present invention provides a deposition type identification model building method, including:

obtaining a sample set according to a plurality of selected logging curves and deposit type data of at least one well, wherein samples in the sample set comprise curve segments and deposit types of the selected logging curves in a certain depth range of the corresponding well;

and setting a network structure and network parameters of the deep neural network according to a set rule, and training the set deep neural network by adopting a periodic attenuation type learning rate through the sample set so as to obtain a deposit type identification model for predicting the target well deposit type data.

In a second aspect, an embodiment of the present invention provides a deposition type identification method, including:

and inputting each selected logging curve of the target well into a deposition type identification model established according to the method, and predicting deposition type data of the target well according to an output result of the model.

In a third aspect, an embodiment of the present invention provides a deposition type identification model building apparatus, including:

the system comprises a sample set establishing module, a sampling module and a sampling module, wherein the sample set establishing module is used for obtaining a sample set according to a plurality of selected logging curves and deposit type data of at least one well, and samples in the sample set comprise curve segments and deposit types of the selected logging curves in a certain depth range corresponding to the well;

the deposition type identification model building module is used for setting the network structure and network parameters of the deep neural network according to the set rule, and training the set deep neural network by adopting the periodic attenuation type learning rate through the sample set so as to obtain a deposition type identification model for predicting the target well deposition type data.

In a fourth aspect, embodiments of the present invention provide a computer program product comprising a computer program/instruction, wherein the computer program/instruction, when executed by a processor, implements the deposition type identification model creation method described above, or implements the deposition type identification method described above.

In a fifth aspect, embodiments of the present disclosure provide a server, comprising: the deposition type identification system comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the deposition type identification model establishment method or the deposition type identification method when executing the program.

The technical scheme provided by the embodiment of the invention has the beneficial effects that at least:

the deposition type identification model establishment method provided by the embodiment of the invention can rapidly and reasonably complete the prediction of the target well deposition type data based on the well data; the periodic attenuation type learning rate is used for optimizing the loss function, so that the convergence speed of the loss function is improved, the loss function has stronger capability of escaping from saddle points and local minimum values, and further the prediction efficiency and the prediction precision are improved.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.

The technical scheme of the invention is further described in detail through the drawings and the embodiments.

Drawings

The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:

FIG. 1 is a flowchart of a deposition type identification model creation method according to a first embodiment of the present invention;

FIG. 2 is a schematic diagram of a network structure of a deep neural network according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a periodic decay learning rate according to an embodiment of the present invention;

FIG. 4 is a flowchart of a specific implementation of training the deep neural network in step S12 of FIG. 1;

FIG. 5 is a flowchart of a method for creating a deposition type identification model according to a second embodiment of the present invention;

FIG. 6 is a graphical representation of visualization of log data in a second embodiment of the present invention;

FIG. 7 is a graph showing various types of log values versus lithology type for a second embodiment of the present invention;

FIG. 8 is a bar chart of the number distribution of label samples according to the second embodiment of the present invention;

FIG. 9 is a graph showing the variation of the loss function/prediction accuracy of the training set and the validation set with training rounds in accordance with the second embodiment of the present invention;

FIG. 10 is a graph comparing a true lithology result with predicted results obtained by various machine learning methods and predicted results obtained by the second embodiment of the present invention;

FIG. 11 is a graph comparing predicted results with real categories using an unsupervised machine learning algorithm, a supervised neural network algorithm (embodiment three), and a semi-supervised algorithm;

fig. 12 is a schematic structural diagram of a deposition type identification model building apparatus according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although only preferred methods and materials are described herein, any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention. All documents mentioned in this specification are incorporated by reference for the purpose of disclosing and describing the methods and/or materials associated with the documents. In case of conflict with any incorporated document, the present specification will control.

In order to solve the problem that the deposition type is difficult to quickly and reasonably identify in the prior art, the embodiment of the invention provides a deposition type identification model establishment method, a deposition type identification method and a related device, and the prediction of target well deposition type data can be quickly and reasonably completed based on well data.

Example 1

An embodiment of the present invention provides a deposition type identification model building method, the flow of which is shown in fig. 1, comprising the following steps:

step S11: a sample set is obtained from a plurality of selected logs of at least one well and the deposit type data.

The samples in the sample set include curve segments and deposit types corresponding to each selected log over a range of depths of the well. The deposit type is a deposit phase type or a lithology type. The deposition type is unique over a range of depths.

Further, the plurality of selected log and deposit type data for the well may include log and partially manually interpreted deposit phase or lithology classification results of natural gamma, resistivity, density porosity, and neutron porosity; the log may also include log obtained by further calculation from conventional log such as longitudinal wave impedance, poisson's ratio, density and longitudinal wave and transverse wave velocity.

Firstly, preprocessing the acquired logging curve, including outlier rejection, normalization, mean value return-to-zero and other processing.

The determining of the selected logging curve type can be that a plurality of logging curves and deposit type data of at least one well are firstly obtained, and the selected logging curve type is screened out through correlation analysis of the logging curve type and the deposit type.

In some embodiments, after obtaining the sample set, determining whether the sample set meets the training number requirement may further include; if not, supplementing the sample set by selecting a machine learning method.

The sample set meets the training quantity requirement, firstly, the total quantity of samples meets the quantity requirement; secondly, the method can further comprise the steps of drawing a distribution histogram of the deposition types contained in the sample set, determining that the number of samples corresponding to each deposition type is approximately equivalent, and if the number of samples corresponding to one or more deposition types is obviously smaller than the number of samples of other deposition types, supplementing the samples of the deposition type with the smaller number of samples.

Step S12: and setting a network structure and network parameters of the deep neural network according to a set rule, and training the set deep neural network by adopting a periodic attenuation type learning rate through a sample set to obtain a deposit type identification model for predicting target well deposit type data.

Setting the network structure of the deep neural network according to a set rule, which can comprise setting the number of neurons at the input end of the deep neural network according to the number of types of the selected logging curves; and setting the number of neurons at the output end of the deep neural network according to the number of types of deposition types contained in the sample set.

More preferably, the number of neurons at the input end of the deep neural network is set to be the same as the number of types of the selected logging curves; the number of neurons at the output end of the deep neural network is set to be the same as the number of types of deposition types contained in the sample set.

Setting network parameters of the deep neural network according to a set rule, which specifically may include:

setting the activation function of the deep neural network as a softmax function:

wherein x is _i Represents the probability of being output as the i-th class, S (x _i ) And outputting an activation function value corresponding to the probability of the ith class, wherein n is the total class number.

The softmax function functions to normalize the output components corresponding to each class so that the sum of the components is 1.

Setting a loss function of the deep neural network as a cross entropy loss function:

wherein J is the cross entropy loss function value, y _i As a label of the i-th class, if the class is i, y _i =1, otherwise equal to 0.

Referring to fig. 2, a schematic diagram of a network structure of a deep neural network is shown; referring to fig. 3, a periodic decay type learning rate is shown, wherein the abscissa is the training round, and the ordinate is the learning rate value corresponding to the training round.

Further, the periodic attenuation type learning rate, namely, the learning rate changes periodically along with the increase of iteration rounds; while each large period comprises several small periods, and the peak value of each small period is in attenuation type change from large to small in one large period.

For the periodic decay learning rate, the number of small cycles included in one large cycle, the width of the small cycle (i.e., the training period covered by the small cycle), the maximum peak value of the small cycle, and the peak decay rate need to be preset.

In some embodiments, after setting the network structure and network parameters of the deep neural network, the training process of the specific model, as shown in fig. 4, may include the following steps:

step S121: the sample set is divided into a training set and a test set.

For example, the training set and the test set may be randomly divided in a ratio of 8:2; alternatively, other ratios are possible.

Step S122: and (3) performing iterative training on the set deep neural network according to the set initial training round by adopting a periodic attenuation type learning rate through the training set, and determining a preferable training round according to the output training log.

The initial training round is often set to be large so that the transition of the validation set loss function from large to small to large can be seen on the curve of the loss function over the rounds.

According to the output training log, the corresponding training round of the training set when the loss function of the verification set just decreases to the minimum along with the increase of the training rounds is used as the optimal training round.

Step S123: and according to the optimal training round, performing iterative training on the set deep neural network by adopting a periodic attenuation type learning rate through a training set to obtain a deposit type identification model for predicting the deposit type data of the target well.

The method for establishing the deposition type identification model provided by the embodiment of the invention can rapidly and reasonably complete the prediction of the target well deposition type data based on the well data; the periodic attenuation type learning rate is used for optimizing the loss function, so that the convergence speed of the loss function is improved, the loss function has stronger capability of escaping from saddle points and local minimum values, and further the prediction efficiency and the prediction precision are improved.

Example two

The second embodiment of the present invention provides a specific implementation flow of a deposition type identification model building method, as shown in fig. 5, including the following steps:

step S51: natural gamma, natural potential, well diameter, shallow\medium\deep resistivity, sonic, neutron porosity, density porosity log and lithology interpretation results are input.

Step S52: and carrying out outlier rejection, mean value zeroing and variance unitizing treatment on the input logging curve.

The average value is zero, namely the average value of the logging curve values after the treatment is zero, so that the data volume and the calculated volume can be reduced, the data management is convenient, and the training and the learning are also convenient; variance unitization is normalization.

The log is processed to facilitate subsequent data visualization, data optimization and model training.

Step S53: visualization of log curves, drawing a crossing diagram of log curve values and deposit types of each type, screening sensitive log curve types, and establishing a sample set for deposit type identification model training.

Log data visualization referring to fig. 6, a log visualization diagram of natural gamma, resistivity, and two types of porosity is shown.

The sedimentary type is exemplified by lithology type, and referring to fig. 7, a cross plot of various types of log values plotted against the sedimentary type is shown; the log is preferably used for subsequent training based on the quality, quantity, and correlation between the data, and the degree of discrimination of the data from the target.

The log used in embodiment two includes: natural gamma, natural potential, well diameter, resistivity, density porosity, neutron porosity, acoustic waves. Through data preprocessing and intersection graph analysis, natural gamma, resistivity, density porosity and neutron porosity are screened out for the input of the neural network. The manually-explained lithology tags corresponding to the input data include 9 kinds of the manually-explained lithology tags, which are respectively: barrier limestone, mudstone limestone, dolomite, mudstone, sea-phase siltstone, river-phase siltstone, beach-phase siltstone, sandstone.

And (3) a bar chart (shown in fig. 8) of the number of the samples of each label is drawn, the labels with the smaller number of the samples are screened, and the samples of the labels are supplemented by using a selected machine learning method, so that a sample set is obtained.

Step S54: the sample set is divided into a training set and a test set.

Step S55: and building a neural network model, and training the neural network model by using a periodic attenuation type learning rate through a training set.

The second embodiment is implemented in the python3.6 language, the neural network structure is built using the PyTorch framework, and the operating system used for training and testing is Windows 10. The machine is configured to: CPU model: intel (R) Xeon (R) Gold 6234CPU@3.30GHz; memory size: 256GB; GPU model: nvidia Quadro RTX 5000 and 5000.

And predicting test set data by using the trained neural network model, and fine-tuning the training turn and the depth and width of the neural network according to the turn change of a loss function convergence curve.

The change of the loss functions of the training set and the verification set along with the training rounds is shown in fig. 9, and the more the training rounds are, the smaller the loss functions are; however, too many training rounds can cause the model to learn over-fitting, which can instead reduce the accuracy of model prediction. It can be determined from fig. 9 that the training round corresponding to the verification set when the loss function just bottoms out is the optimal training round.

Fig. 9 also verifies that the improvement of training efficiency by the periodic decay rate allows convergence to be achieved quickly, and the training duration of the second embodiment is reduced by 60% when the same 80% prediction accuracy is obtained as compared to neural network training of constant or decay array learning rates.

Fig. 10 is a comparison of a true lithology result with a predicted result obtained through various machine learning methods and a predicted result obtained through deep learning. The prediction precision obtained by the support vector machine and the deep learning is higher than that of other methods, the prediction precision of sandstone and sea siltstone is higher, and the number of the original label samples corresponding to the support vector machine and the deep learning is insufficient.

Example III

The third embodiment of the invention provides an application example of a deposition type identification model building method.

And (3) taking only one well as a known well, simulating the condition of insufficient sample quantity of the label, and taking 5 curves as input, wherein the sample label comprises sandstone and shale. Through model training of the well, classification prediction is carried out on another well, an experimental result shows (as shown in fig. 11), the semi-supervised learning prediction effect taking the result after machine learning classification as input has a certain gap from the supervised learning represented by deep learning, but the problem of insufficient label samples is greatly relieved, and the capability of solving the problem of small samples is verified.

Based on the inventive concept, the embodiment of the invention further provides a deposition type identification method, which comprises the following steps:

and inputting each selected logging curve of the target well into a deposition type identification model established according to the method, and predicting deposition type data of the target well according to an output result of the model.

Based on the inventive concept, the embodiment of the invention further provides a deposition type identification model building device, the structure of which is shown in fig. 12, including:

a sample set establishing module 121, configured to obtain a sample set according to a plurality of selected logging curves and deposition type data of at least one well, where samples in the sample set include curve segments and deposition types corresponding to each selected logging curve within a certain depth range of the well;

the deposition type identification model building module 122 is configured to set a network structure and network parameters of the deep neural network according to a set rule, and train the set deep neural network by using a periodic attenuation type learning rate through the sample set, so as to obtain a deposition type identification model for predicting the deposition type data of the target well.

The specific manner in which the various modules perform the operations in the apparatus of the above embodiments have been described in detail in connection with the embodiments of the method, and will not be described in detail herein.

Based on the inventive concept, the embodiments of the present invention further provide a computer program product, including a computer program/instruction, where the computer program/instruction implements the deposition type identification model building method or implements the deposition type identification method when executed by a processor.

Unless specifically stated otherwise, terms such as processing, computing, calculating, determining, displaying, or the like, may refer to an action and/or process of one or more processing or computing systems, or similar devices, that manipulates and transforms data represented as physical (e.g., electronic) quantities within the processing system's registers or memories into other data similarly represented as physical quantities within the processing system's memories, registers or other such information storage, transmission or display devices. Information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that may be referenced throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

It should be understood that the specific order or hierarchy of steps in the processes disclosed are examples of exemplary approaches. Based on design preferences, it is understood that the specific order or hierarchy of steps in the processes may be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and are not meant to be limited to the specific order or hierarchy presented.

In the foregoing detailed description, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments of the subject matter require more features than are expressly recited in each claim. Rather, as the following claims reflect, invention lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate preferred embodiment of this invention.

Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. The processor and the storage medium may reside as discrete components in a user terminal.

For a software implementation, the techniques described herein may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. These software codes may be stored in memory units and executed by processors. The memory unit may be implemented within the processor or external to the processor, in which case it can be communicatively coupled to the processor via various means as is known in the art.

The foregoing description includes examples of one or more embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the aforementioned embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations of various embodiments are possible. Accordingly, the embodiments described herein are intended to embrace all such alterations, modifications and variations that fall within the scope of the appended claims. Furthermore, as used in the specification or claims, the term "comprising" is intended to be inclusive in a manner similar to the term "comprising," as interpreted when employed as a transitional word in a claim. Furthermore, any use of the term "or" in the specification of the claims is intended to mean "non-exclusive or".

Claims (13)

1. A deposition type identification model building method, characterized by comprising:

obtaining a sample set according to a plurality of selected logging curves and deposit type data of at least one well, wherein samples in the sample set comprise curve segments and deposit types of the selected logging curves in a certain depth range of the corresponding well;

and setting a network structure and network parameters of the deep neural network according to a set rule, and training the set deep neural network by adopting a periodic attenuation type learning rate through the sample set so as to obtain a deposit type identification model for predicting the target well deposit type data.

2. The method of claim 1, wherein the setting the network structure of the deep neural network according to the set rule specifically comprises:

setting the number of neurons at the input end of the deep neural network according to the number of types of the selected logging curves;

and setting the number of neurons at the output end of the deep neural network according to the number of types of deposition types contained in the sample set.

3. The method of claim 2, wherein the setting the network structure of the deep neural network according to the set rule specifically comprises:

setting the same number of neurons at the input end of the deep neural network as the number of types of the selected logging curves;

the number of neurons at the output end of the deep neural network is set to be the same as the number of types of deposition types contained in the sample set.

4. The method of claim 1, wherein setting network parameters of the deep neural network according to a set rule, specifically comprises:

setting the activation function of the deep neural network as a softmax function:

wherein x is _i Represents the probability of being output as the i-th class, S (x _i ) For outputting an activation function value corresponding to the probability of the ith class, n is the total class number;

setting a loss function of the deep neural network as a cross entropy loss function:

wherein J is the cross entropy loss function value, y _i Is a class i tag.

5. The method of claim 1, wherein training the set-up deep neural network with a periodically decaying learning rate through the sample set to obtain a deposit type recognition model for predicting the well deposit type data of interest, specifically comprises:

dividing the sample set into a training set and a testing set;

through the training set, adopting a periodic attenuation type learning rate, performing iterative training on the set deep neural network according to a set initial training round, and determining a preferable training round according to an output training log;

and according to the optimal training round, performing iterative training on the set depth neural network by adopting the periodic attenuation type learning rate through the training set so as to obtain a deposit type identification model for predicting the target well deposit type data.

6. The method of claim 1, wherein the obtaining a sample set from the plurality of selected logs and the deposition type data for the at least one well further comprises:

judging whether the sample set meets the requirement of training quantity or not;

if not, supplementing the sample set by selecting a machine learning method.

7. The method of claim 1, wherein prior to obtaining the sample set from the plurality of selected logs and the deposition type data for the at least one well, further comprising:

and screening out the selected logging curve type through correlation analysis of the logging curve type and the deposition type according to the logging curve and the deposition type data of at least one well.

8. The method of any one of claims 1-7, wherein the selected log is a normalized log.

9. The method of claim 8, wherein the selected log is a mean zeroed log.

10. The method of any one of claims 1 to 7, wherein the deposit type is a deposit phase type or a lithology type.

11. A deposition type identification method, comprising:

inputting each selected logging curve of a target well into a deposition type identification model established according to the method of any one of claims 1-10, and predicting deposition type data of the target well according to the output result of the model.

12. A deposition type identification model creation apparatus, characterized by comprising:

the system comprises a sample set establishing module, a sampling module and a sampling module, wherein the sample set establishing module is used for obtaining a sample set according to a plurality of selected logging curves and deposit type data of at least one well, and samples in the sample set comprise curve segments and deposit types of the selected logging curves in a certain depth range corresponding to the well;

the deposition type identification model building module is used for setting the network structure and network parameters of the deep neural network according to the set rule, and training the set deep neural network by adopting the periodic attenuation type learning rate through the sample set so as to obtain a deposition type identification model for predicting the target well deposition type data.

13. A computer program product comprising computer program/instructions which, when executed by a processor, implements the deposition type identification model building method according to any one of claims 1 to 10, or implements the deposition type identification method according to claim 11.