Seismic data processing method and device and electronic equipment

Abstract

The embodiment of the invention provides a seismic data processing method and device and electronic equipment. The method is applied to a main operation node of a seismic data processing system for carrying out parallel processing on seismic data. Receiving first seismic data and service information using the first seismic data; performing space-time slicing on the first seismic data to obtain a plurality of first sub-data blocks which are sequentially arranged in a space-time dimension; sending the plurality of first sub-data blocks to respective corresponding sub-operation nodes, and reasoning the plurality of sampling data of the first sub-data blocks in parallel by respective data processing models of the plurality of sub-operation nodes to obtain a plurality of reasoning results; and splicing the plurality of reasoning results according to the arrangement order of the respective first sub-data blocks to obtain a target result. The method is used for achieving the effect of improving the seismic data processing efficiency in the reasoning stage.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to a seismic data processing method, device and electronic equipment. Background Art

[0002] With the development of science and technology, different fields have higher and higher requirements for the efficiency and accuracy of large-scale data processing. Since artificial intelligence technology has strong adaptive learning ability and high-precision prediction ability, it has been applied to many fields. Geologists can extract and interpret underground geological structures and characteristics by analyzing the collected seismic data, which can be applied to resource assessment, geological risk assessment, resource exploration and other scenarios. Therefore, artificial intelligence technology is also used to process seismic data.

[0003] Artificial intelligence operations are divided into two stages: training and reasoning. In the training stage, a data model is generated by training a large amount of data. In the reasoning stage, the trained data model is deployed to the relevant environment, and the data model is used to perform operations on the input data to obtain the reasoning result. In the training stage, due to the large amount of data used, a large amount of computing resources are consumed. Therefore, a parallel data processing solution is usually used for model training. In the reasoning stage, since the amount of data processed by the data model is relatively small, the relevant technology uses a serialized data processing solution to process the input data.

[0004] For seismic data processing scenarios, due to the fact that seismic data has the characteristic of extremely large single data text, there is a technical problem of low efficiency when using serialized data processing solutions to process seismic data in the inference stage. Summary of the invention

[0005] The embodiments of the present application provide a seismic data processing method, device and electronic device to improve the processing efficiency of seismic data in the reasoning stage.

[0006] In a first aspect, an embodiment of the present application provides a seismic data processing method, which is applied to a main operation node of a seismic data processing system for parallel processing of seismic data, wherein the seismic data processing system includes the main operation node and a plurality of sub-operation nodes; the method includes:

[0007] Receiving first seismic data and service information using the first seismic data; the first seismic data includes a plurality of sampling data, and the service information indicates a target service;

[0008] Slicing the first seismic data in time and space to obtain a plurality of first sub-data blocks sequentially arranged in time and space dimensions, wherein each first sub-data block includes a plurality of sampling data;

[0009] Sending the plurality of first sub-data blocks to the respective corresponding sub-operation nodes, and using the respective data processing models of the plurality of sub-operation nodes to infer the plurality of sampled data of the first sub-data blocks in parallel to obtain inference results respectively corresponding to the plurality of first sub-data blocks;

[0010] A plurality of the inference results are obtained from a plurality of the sub-operation nodes, and the plurality of the inference results are spliced ​​according to the arrangement order of the respective first sub-data blocks to obtain a target result.

[0011] In a possible implementation, there is repeated first sampling data between every two adjacent first sub-data blocks, and the inference result corresponding to each first sub-data block includes a second sub-data block, and the second sub-data block includes a plurality of target data, and each target data corresponds one-to-one to each first sampling data of the first sub-data block; and the acquiring of the plurality of inference results from the plurality of sub-operation nodes and splicing of the plurality of inference results according to the arrangement order of the first sub-data blocks includes:

[0012] The plurality of second sub-data blocks are concatenated according to their respective arrangement orders to obtain a target result; wherein the second sub-data block inherits the arrangement order of the corresponding first sub-data block; wherein,

[0013] During the process of splicing multiple second sub-data blocks, for every two adjacent second sub-data blocks, the first target data of each of the two second sub-data blocks are fused based on the signal-to-noise ratio of each of the two second sub-data blocks, wherein each first target data corresponds to a first sampling data.

[0014] In a possible implementation manner, the step of performing spatiotemporal slicing on the first seismic data to obtain a plurality of first sub-data blocks sequentially arranged in the spatiotemporal dimension includes:

[0015] Determine the minimum space-time range corresponding to the plurality of sampling data;

[0016] Based on the number of the sub-operation nodes, the minimum spatiotemporal range is sliced ​​in an overlapping manner to obtain a plurality of sub-ranges, wherein there is an overlapping sub-range between every two adjacent sub-ranges;

[0017] For each sub-range, a plurality of sampling data belonging to the sub-range is acquired from the first seismic data, and the plurality of sampling data is used as a first sub-data block;

[0018] For every two adjacent sub-ranges, the first sub-data blocks respectively corresponding to the two sub-ranges are two adjacent first sub-data blocks, and the sampling data within the overlapping sub-range of the two sub-ranges are the first sampling data of the two adjacent first sub-data blocks.

[0019] In a possible implementation, the method further includes:

[0020] Obtaining an identifier of each sub-computing node, and establishing an association relationship between the identifier and a first sub-data block to be sent to the sub-computing node;

[0021] The sending of the plurality of first sub-data blocks to the corresponding sub-operation nodes comprises:

[0022] For each first sub-data block, the first sub-data block is sent to a corresponding sub-operation node according to the association relationship.

[0023] In a possible implementation manner, the step of fusing the plurality of repeated target data based on the signal-to-noise ratios of the two second sub-data blocks for the first target data in each of two adjacent second sub-data blocks includes:

[0024] Obtaining a signal-to-noise ratio of each of the two second sub-data blocks;

[0025] Determining respective weights of the two second sub-data blocks according to the respective signal-to-noise ratios of the two second sub-data blocks;

[0026] The first target data of each of the two second sub-data blocks are weightedly summed according to the weights to obtain fused data.

[0027] In one possible implementation, the main computing node and the multiple sub-computing nodes are deployed based on an independent node cluster, or the main computing node and the multiple sub-computing nodes are deployed based on a K8s cluster, and the data processing model is loaded on each sub-computing node based on the Kubeflow management platform to process the first sub-data block input therein.

[0028] In a second aspect, an embodiment of the present application provides a seismic data processing device, which is applied to a main operation node of a seismic data processing system for parallel processing of seismic data, wherein the seismic data processing system includes the main operation node and a plurality of sub-operation nodes; the device includes:

[0029] An input module, configured to receive first seismic data and business information using the first seismic data; the first seismic data includes a plurality of sampled data, and the business information indicates a target business;

[0030] A slicing module, configured to slice the first seismic data in time and space to obtain a plurality of first sub-data blocks sequentially arranged in time and space dimensions, wherein each first sub-data block includes a plurality of sample data;

[0031] A computing module, used for sending the plurality of first sub-data blocks to the corresponding sub-computing nodes respectively, and for the data processing models of the plurality of sub-computing nodes respectively to infer the plurality of sampled data of the first sub-data blocks in parallel, to obtain inference results respectively corresponding to the plurality of first sub-data blocks;

[0032] A splicing module is used to obtain multiple reasoning results from multiple sub-operation nodes, and splice the multiple reasoning results according to the arrangement order of the first sub-data blocks to obtain a target result.

[0033] In a third aspect, an embodiment of the present application provides a seismic data processing device, including: a memory, a processor;

[0034] The memory stores computer-executable instructions;

[0035] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementations of the first aspect.

[0037] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0038] The seismic data processing method, device and electronic device provided in the embodiments of the present application slice the seismic data into multiple independent sub-data blocks in the inference stage, and use multiple sub-operation nodes to process the multiple sub-data blocks in parallel to obtain multiple inference results. The main operation node collects multiple inference results and splices the multiple inference results based on the arrangement order of the multiple sub-data blocks to obtain the target result. Since the inference stage uses multiple computing resources to process the seismic data in parallel, the processing efficiency of the seismic data in the inference stage is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0040] Figure 1 A schematic diagram of the flow chart of the seismic data processing method provided in this application;

[0041] Figure 2 A schematic diagram of the structure of the seismic data processing device provided in this application;

[0042] Figure 3 A schematic diagram of the structure of the seismic data processing equipment provided in this application.

[0043] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0044] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0045] First, the terms involved in this application are explained:

[0046] GPU: Graphics Processing Unit, a microprocessor that is specialized in performing image and graphics-related calculations on personal computers, workstations, game consoles, and some mobile devices.

[0047] k8s cluster: refers to the Kubernetes cluster, which is an open source container orchestration platform for automating the deployment, expansion and management of containerized applications. The k8s cluster consists of multiple nodes, including master nodes and worker nodes. The master node is responsible for managing the status and configuration information of the entire cluster, and the worker node is used to run container applications.

[0048] With the development of science and technology, different fields have higher and higher requirements for the efficiency and accuracy of large-scale data processing. Since artificial intelligence technology has strong adaptive learning ability and high-precision prediction ability, it has been applied to many fields. Geologists can extract and interpret underground geological structures and characteristics by analyzing the collected seismic data, which can be applied to resource assessment, geological risk assessment, resource exploration and other scenarios. Therefore, artificial intelligence technology is also used to process seismic data.

[0049] Seismic data in SEGY format is a type of data widely used in the seismic exploration industry. It can effectively store a large amount of seismic waveform data and has the characteristic of a single data file being extremely large. SEGY files usually consist of three parts: a text header, a binary header, and multiple seismic traces. Among them, the text header is used to store the basic information and description of the file. The binary header is used to describe the overall characteristics of the data. Multiple seismic traces store actual seismic data, and each seismic trace includes trace header information and seismic trace data. The trace header information generally stores information such as the line number, trace number, number of sampling points, and geodetic coordinates corresponding to the seismic trace. Seismic trace data is the data obtained by sampling the waveform of the seismic signal at a certain time interval, and then recording this series of discrete amplitude values ​​in some way.

[0050] Multiple sampling data from multiple seismic channels map out the subtle ups and downs of the various stratum interfaces and the complex changes in the stratum structure under a unified reference plane. By slicing these data, a series of sections reflecting the continuity and change characteristics of the strata in a certain direction can be obtained. These sections are not only a static snapshot of the strata in a certain direction, but also a record of the dynamic evolution of the underground geological structure. Therefore, the data of a single or multiple slices can be analyzed to analyze the geological structure, which can be used to evaluate the potential of mineral resources, plan oil and gas development plans, and conduct geological engineering risk assessments.

[0051] Artificial intelligence operations are divided into two stages: training and reasoning. In the training stage, a data model is generated by training a large amount of data. In the reasoning stage, the trained data model is deployed to the relevant environment, and the data model is used to perform operations on the input data to obtain the reasoning result. In the training stage, due to the large amount of data used, a large amount of computing resources are consumed. Therefore, a parallel data processing solution is usually used for model training. In the reasoning stage, since the amount of data processed by the data model is relatively small, the relevant technology uses a serialized data processing solution to process the input data.

[0052] Based on the above analysis, it can be seen that for seismic data processing scenarios, due to the characteristics of seismic data having extremely large single data text, there is a technical problem of low efficiency when using a serialized data processing solution to process seismic data in the inference stage.

[0053] The seismic data processing method, device and electronic device provided by the present application slice the seismic data into multiple independent sub-data blocks during the inference stage, and use multiple sub-operation nodes to process the multiple sub-data blocks in parallel to obtain multiple inference results. The main operation node collects multiple inference results and splices the multiple inference results based on the arrangement order of the multiple sub-data blocks to obtain the target result. At the same time, by slicing the seismic data in an overlapping manner and splicing the multiple inference results based on the signal-to-noise ratio, the problem of edge splicing anomalies caused by data model processing is reduced. Since the seismic data is processed in parallel using multiple computing resources during the inference stage, the processing efficiency of the seismic data during the inference stage is improved. At the same time, since the seismic data is sliced ​​in an overlapping manner, the accuracy of the target result obtained by seismic data processing is improved.

[0054] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0055] Figure 1 A flow chart of the seismic data processing method provided in this application is as follows: Figure 1 As shown, the present application is applied to a main operation node of a seismic data processing system for parallel processing of seismic data. The seismic data processing system includes a main operation node and multiple sub-operation nodes. The method provided by the present application includes:

[0056] S110: Receive first seismic data and business information using the first seismic data; the first seismic data includes a plurality of sampling data, and the business information indicates a target business.

[0057] In this step, the first seismic data is in the data format of SEGY, including multiple seismic traces, each of which includes trace header information and seismic trace data. The seismic trace data is obtained by sampling the waveform of the seismic signal at a certain time interval, and then recording this series of discrete amplitudes in a certain way. Therefore, each seismic trace data contains multiple discrete sampling data, and multiple sampling data of multiple seismic traces are combined into multiple sampling data of the first seismic data.

[0058] The business information determines the business scenario for using the first seismic data, and the business goal determines the seismic information that the target result needs to carry after processing the first seismic data.

[0059] S120: Slice the first seismic data in time and space to obtain a plurality of first sub-data blocks sequentially arranged in the time and space dimensions, wherein each first sub-data block includes a plurality of sampling data.

[0060] In this step, slicing the first seismic data in time and space refers to slicing the first seismic data in a temporal or spatial order.

[0061] For each seismic trace data of the first seismic data, all sampling data within a complete sampling time range from the start of sampling to the end of sampling are recorded. Within this range, sampling is performed at a certain time interval, thereby obtaining multiple sampling data. The sampling time information is recorded in the trace header information of each seismic trace.

[0062] For each seismic trace of the first seismic data, a main survey line number and a connecting survey line number are stored in the trace header information. The main survey line numbers of multiple seismic traces are combined to obtain a main survey line, and the connecting survey line numbers of multiple seismic traces are combined to obtain a connecting survey line. In seismic exploration, survey lines are arranged in a certain direction to record the propagation of seismic waves. The main survey line is the main or basic survey line among these survey lines. The connecting survey line is a survey line that intersects or is perpendicular to the main survey line. The main survey line and the connecting survey line are used to form a two-dimensional coordinate system, which is used to replace the actual sampling coordinate system in the field. Therefore, the main survey line number and the connecting survey line number can be used to represent the spatial information of the sampling data.

[0063] According to the three-dimensional coordinate system composed of time, main survey line number and contact survey line number, multiple sampling data of the first seismic data can be mapped into a three-dimensional space. Each sampling data corresponds to a data point in the three-dimensional space, and all sampling data form a three-dimensional array.

[0064] Slicing the first earthquake data according to time means slicing multiple data points in three-dimensional space in a direction perpendicular to the time axis to obtain multiple first sub-data blocks; slicing the first earthquake data according to space means slicing multiple data points in three-dimensional space in a direction perpendicular to the main survey line number coordinate axis or the contact survey line number coordinate axis to obtain multiple first sub-data blocks. Each first sub-data block corresponds to a three-dimensional array.

[0065] It should be noted that the information composed of the multiple sampling data of the multiple first sub-data blocks does not include the file header information, binary header information and trace header information of each seismic trace of the first seismic data.

[0066] S130: Send the multiple first sub-data blocks to the corresponding sub-operation nodes respectively, and use the data processing models of the multiple sub-operation nodes to infer the multiple sampling data of the first sub-data blocks in parallel to obtain the inference results corresponding to the multiple first sub-data blocks respectively.

[0067] In this step, the main computing node and multiple sub-computing nodes form a node cluster, and each computing node corresponds to a GPU. Data can be exchanged between the main computing node and multiple sub-computing nodes, and the nodes communicate data and cooperate with each other through the network. Each sub-computing node uses its own GPU to perform data calculations. During the parallel reasoning process of multiple sub-computing nodes, there is no data communication between sub-computing nodes.

[0068] When each sub-operation node uses the data model to infer the first sub-data block, it actually performs data operation on the three-dimensional array corresponding to each first sub-data block.

[0069] S140: Acquire multiple inference results from multiple sub-operation nodes, and concatenate the multiple inference results according to the arrangement order of the respective first sub-data blocks to obtain a target result.

[0070] The seismic data processing method provided in the embodiment of the present application slices the seismic data into multiple independent sub-data blocks in the reasoning stage, and uses multiple sub-operation nodes to process the multiple sub-data blocks in parallel to obtain multiple reasoning results. The main operation node collects multiple reasoning results and splices the multiple reasoning results based on the arrangement order of the multiple sub-data blocks to obtain the target result. Since the reasoning stage uses multiple computing resources to process the seismic data in parallel, the processing efficiency of the seismic data in the reasoning stage is improved.

[0071] The following takes the processing of seismic data in a specific SEGY format as an example to explain in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems.

[0072] In a specific implementation, for the multiple first sub-data blocks obtained by slicing in step S120 and the multiple inference results obtained by inference in step S130, considering that in actual applications, after the first sub-data blocks are processed using a data processing model, the inference results obtained are prone to abnormal data at the edges, so every two adjacent first sub-data blocks are processed to have overlapping parts. The specific implementation is as follows:

[0073] There is repeated first sampling data between every two adjacent first sub-data blocks. The inference result corresponding to each first sub-data block includes a second sub-data block. The second sub-data block contains multiple target data. Each target data corresponds to each sampling data of the first sub-data block one by one.

[0074] In step S140, multiple inference results are obtained from multiple sub-operation nodes, and the multiple inference results are spliced ​​according to the arrangement order of each first sub-data block, including: splicing multiple second sub-data blocks according to their respective arrangement orders to obtain a target result; wherein the second sub-data block inherits the arrangement order of the corresponding first sub-data block.

[0075] During the splicing of multiple second sub-data blocks, for every two adjacent second sub-data blocks, the first target data of the two second sub-data blocks are fused based on the signal-to-noise ratio of the two second sub-data blocks, wherein each first target data corresponds to a first sampling data.

[0076] It should be noted that when each sub-operation node uses the data model to infer the first sub-data block, it actually performs data operations on the three-dimensional array corresponding to each first sub-data block. The corresponding second sub-data block is also a three-dimensional array during the operation. Therefore, after completing the splicing of multiple second sub-data blocks, a complete three-dimensional array is obtained. It is necessary to fill the multiple data points in the complete three-dimensional array into the corresponding seismic channel data according to their respective time coordinates, main survey line number coordinates, and contact survey line number coordinates for recombining, and fill in the header information, binary header information, and text header information according to the actual situation of multiple target data to obtain a target result in SEGY data format.

[0077] In some implementations of this embodiment, the step S120 of performing spatiotemporal slicing on the first seismic data to obtain a plurality of first sub-data blocks sequentially arranged in the spatiotemporal dimension includes the following sub-steps:

[0078] S121: Determine the minimum space-time range corresponding to the multiple sampling data.

[0079] S122: Based on the number of sub-operation nodes, the minimum spatiotemporal range is sliced ​​in an overlapping manner to obtain a plurality of sub-ranges, wherein there is an overlapping sub-range between every two adjacent sub-ranges.

[0080] S123: For each sub-range, obtain a plurality of sampling data belonging to the sub-range in the first seismic data, and use the plurality of sampling data as a first sub-data block.

[0081] For every two adjacent sub-ranges, the first sub-data blocks respectively corresponding to the two sub-ranges are two adjacent first sub-data blocks, and the sampling data in the overlapping sub-range of the two sub-ranges are the first sampling data of the two adjacent first sub-data blocks.

[0082] For example, if it is determined based on the target business that the first seismic data is currently sliced ​​in a direction perpendicular to the coordinate axis of the contact line number, and it is determined that the contact line range to be sliced ​​is 1 meter to 1000 meters, the total number of sub-operation nodes is 10. It is determined that the contact line range of 1 meter to 1000 meters is divided into 10 sub-ranges, that is, the minimum sub-range is 100 meters.

[0083] If the overlapping part is set to 5 meters, the sub-ranges of the contact survey line range are: 1 meter to 105 meters, 100 meters to 205 meters, 200 meters to 305, and so on. The last three sub-ranges are 700 meters to 805 meters, 800 meters to 905 meters, and 900 meters to 1000 meters.

[0084] For a first sub-data block composed of a plurality of sampling data within 1 meter to 105 meters and a first sub-data block composed of a plurality of sampling data within 100 meters to 205 meters in the first seismic data, the two first sub-data blocks are adjacent first sub-data blocks. 100 meters to 105 meters is an overlapping sub-range, and the plurality of sampling data within the overlapping sub-range constitutes a plurality of repeated first sampling data.

[0085] Through the above-mentioned overlapping slicing process of two adjacent first sub-data blocks, the abnormal values ​​of multiple target data in the second sub-data block are reduced, thereby ensuring that the target result obtained by splicing is more accurate.

[0086] In some implementations of this embodiment, a data identifier is used to ensure that multiple inference results are spliced ​​in the correct order, specifically including:

[0087] An identifier of each sub-operation node is obtained, and an association relationship is established between the identifier and the first sub-data block to be sent to the sub-operation node.

[0088] Optionally, the obtained identifier of the sub-computing node is the GPU number of the sub-computing node. The GPU number is generally fixed, and is composed of a series of numbers, letters, or a combination thereof, and is unique. Therefore, it can be used as a basis for identifying the first sub-data block.

[0089] Sending the plurality of first sub-data blocks to the corresponding sub-operation nodes includes:

[0090] For each first sub-data block, the first sub-data block is sent to a corresponding sub-operation node according to the association relationship.

[0091] The second sub-data block inherits the arrangement order of the first sub-data block by establishing an association relationship between the identifier of the sub-operation node and the second sub-data block obtained by the sub-operation node. This makes it convenient for the main operation node to match the identifier of the second sub-data block with that of the first sub-data block, thereby determining the splicing order of the second sub-data block according to the splicing order of the first sub-data block.

[0092] In some implementations of this embodiment, the first target data in each two adjacent second sub-data blocks are merged based on the signal-to-noise ratios of the two second sub-data blocks, and the multiple repeated target data include the following sub-steps:

[0093] The signal-to-noise ratio of each of the two second sub-data blocks is obtained.

[0094] The respective weights of the two second sub-data blocks are determined according to the respective signal-to-noise ratios of the two second sub-data blocks.

[0095] The first target data of each of the two second sub-data blocks are weightedly summed according to the weights to obtain fused data.

[0096] For the two second sub-data blocks, the greater the signal-to-noise ratio is, the greater the corresponding weight is.

[0097] It should be noted that the two second sub-data blocks may be subjected to noise reduction processing first, and then the effective signal and noise data may be used to calculate the signal-to-noise ratio of each of the two second sub-data blocks. Other methods of calculating the signal-to-noise ratio are also applicable to the technical solution provided in this application, and the specific method of calculating the signal-to-noise ratio may be determined according to actual conditions, and this application does not limit this.

[0098] The signal-to-noise ratio is an important indicator for measuring the signal quality in seismic data. It measures the relative size of the signal strength and the background noise. By assigning a larger weight to the second sub-data block with a large signal-to-noise ratio, more effective information in the overlapping part of the two second sub-data blocks is retained, thereby ensuring that the target result obtained by splicing is more accurate.

[0099] In some implementations of this embodiment, a main computing node and multiple sub-computing nodes are deployed based on an independent node cluster, or a main computing node and multiple sub-computing nodes are deployed based on a K8s cluster, and a data processing model is loaded on each sub-computing node based on the Kubeflow management platform to process the first sub-data block input therein.

[0100] Among them, the Kubeflow management platform is based on the container orchestration and resource management functions of Kubernetes. By splitting the machine learning workflow into a series of containerized tasks, it uses Kubernetes' automatic expansion, fault tolerance and scheduling functions to ensure the efficient execution of machine learning tasks.

[0101] Figure 2 The structural diagram of the seismic data processing device provided in this application is as follows: Figure 2 As shown, the seismic data processing device 20 provided in this embodiment is applied to a main operation node of a seismic data processing system for parallel processing of seismic data. The seismic data processing system includes a main operation node and a plurality of sub-operation nodes. The seismic data processing device 20 includes:

[0102] The input module 201 is used to receive first seismic data and business information using the first seismic data; the first seismic data includes a plurality of sampling data, and the business information indicates a target business.

[0103] The slicing module 202 is used to slice the first seismic data in time and space to obtain a plurality of first sub-data blocks arranged in sequence in the time and space dimensions, wherein each first sub-data block includes a plurality of sampling data.

[0104] The operation module 203 is used to send multiple first sub-data blocks to their respective corresponding sub-operation nodes, and the data processing models of the multiple sub-operation nodes respectively infer the multiple sampling data of the first sub-data blocks in parallel to obtain the inference results corresponding to the multiple first sub-data blocks.

[0105] The splicing module 204 is used to obtain multiple inference results from multiple sub-operation nodes, and splice the multiple inference results according to the arrangement order of the respective first sub-data blocks to obtain a target result.

[0106] In a specific embodiment, the slicing module 202 and the splicing module 204 are also used for having repeated first sampling data between every two adjacent first sub-data blocks, and the inference result corresponding to each first sub-data block includes a second sub-data block, the second sub-data block contains multiple target data, and each target data corresponds one-to-one to each sampling data of the first sub-data block; the multiple second sub-data blocks are spliced ​​according to their respective arrangement orders to obtain the target result; wherein the second sub-data blocks inherit the arrangement order of the corresponding first sub-data blocks; wherein, in the process of splicing the multiple second sub-data blocks, for every two adjacent second sub-data blocks, the first target data of the two second sub-data blocks are fused based on the respective signal-to-noise ratios of the two second sub-data blocks, wherein each first target data corresponds to a first sampling data.

[0107] In some implementations of the present embodiment, the slicing module 202 is also used to determine the minimum space-time range corresponding to multiple sampling data; based on the number of sub-operation nodes, the minimum space-time range is sliced ​​in an overlapping manner to obtain multiple sub-ranges, wherein there is an overlapping sub-range between every two adjacent sub-ranges; for each sub-range, multiple sampling data belonging to the sub-range are obtained in the first seismic data, and the multiple sampling data are used as a first sub-data block; wherein, for every two adjacent sub-ranges, the first sub-data blocks corresponding to the two sub-ranges are two adjacent first sub-data blocks, and the sampling data within the overlapping sub-range of the two sub-ranges is the first sampling data of the two adjacent first sub-data blocks.

[0108] In some implementations of this embodiment, the operation module 203 is also used to obtain the identifier of each sub-operation node, and establish an association relationship between the identifier and the first sub-data block to be sent to the sub-operation node; for each first sub-data block, the first sub-data block is sent to the corresponding sub-operation node according to the association relationship.

[0109] In some implementations of this embodiment, the splicing module 204 is further used to obtain the signal-to-noise ratio of each of the two second sub-data blocks; determine the weights of the two second sub-data blocks according to the signal-to-noise ratio of each of the two second sub-data blocks; and perform weighted summation of the first target data of each of the two second sub-data blocks according to the weights to obtain fused data.

[0110] In some implementations of this embodiment, the input module 201 is also used to deploy a main computing node and multiple sub-computing nodes based on an independent node cluster, or to deploy a main computing node and multiple sub-computing nodes based on a K8s cluster, and to load a data processing model on each sub-computing node based on the Kubeflow management platform to process the first sub-data block input therein.

[0111] The seismic data processing device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and this embodiment will not be described in detail here.

[0112] Figure 3 This is a schematic diagram of the structure of the seismic data processing equipment provided in this application. Figure 3 As shown, the electronic device 30 provided in this embodiment includes: at least one processor 301 and a memory 302. Optionally, the device 30 also includes a communication component 303. The processor 301, the memory 302 and the communication component 303 are connected via a bus.

[0113] In a specific implementation process, at least one processor 301 executes the computer-executable instructions stored in the memory 302, so that at least one processor 301 executes the above method.

[0114] The specific implementation process of the processor 301 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0115] In the above embodiments, it should be understood that the processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the invention can be directly implemented as a hardware processor, or can be implemented by a combination of hardware and software modules in the processor.

[0116] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (NVM), such as at least one disk storage.

[0117] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0118] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0119] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0120] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0121] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0122] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0123] 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 distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0124] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0125] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0126] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0127] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for processing seismic data, characterized in that: A main operation node of a seismic data processing system for parallel processing of seismic data, the seismic data processing system comprising the main operation node and a plurality of sub-operation nodes; the method comprising: Receiving first seismic data and service information using the first seismic data; the first seismic data includes a plurality of sampling data, and the service information indicates a target service; Slicing the first seismic data in time and space to obtain a plurality of first sub-data blocks sequentially arranged in time and space dimensions, wherein each first sub-data block includes a plurality of sampling data; Sending the plurality of first sub-data blocks to the respective corresponding sub-operation nodes, and using the respective data processing models of the plurality of sub-operation nodes to infer the plurality of sampled data of the first sub-data blocks in parallel to obtain inference results respectively corresponding to the plurality of first sub-data blocks; A plurality of the inference results are obtained from a plurality of the sub-operation nodes, and the plurality of the inference results are spliced ​​according to the arrangement order of the respective first sub-data blocks to obtain a target result.

2. The method according to claim 1, characterized in that There is repeated first sampled data between every two adjacent first sub-data blocks, and the inference result corresponding to each first sub-data block includes a second sub-data block, the second sub-data block includes a plurality of target data, and each target data corresponds to each sampled data of the first sub-data block one by one; And the step of obtaining a plurality of the inference results from the plurality of the sub-operation nodes and splicing the plurality of the inference results according to the arrangement order of the first sub-data blocks comprises: The plurality of second sub-data blocks are concatenated according to their respective arrangement orders to obtain a target result; wherein the second sub-data block inherits the arrangement order of the corresponding first sub-data block; wherein, During the process of splicing multiple second sub-data blocks, for every two adjacent second sub-data blocks, the first target data of each of the two second sub-data blocks are fused based on the signal-to-noise ratio of each of the two second sub-data blocks, wherein each first target data corresponds to a first sampling data.

3. The method according to claim 2, characterized in that The step of slicing the first seismic data in time and space to obtain a plurality of first sub-data blocks sequentially arranged in time and space dimensions includes: Determine the minimum space-time range corresponding to the plurality of sampling data; Based on the number of the sub-operation nodes, the minimum spatiotemporal range is sliced ​​in an overlapping manner to obtain a plurality of sub-ranges, wherein there is an overlapping sub-range between every two adjacent sub-ranges; For each sub-range, a plurality of sampling data belonging to the sub-range is acquired from the first seismic data, and the plurality of sampling data is used as a first sub-data block; For every two adjacent sub-ranges, the first sub-data blocks respectively corresponding to the two sub-ranges are two adjacent first sub-data blocks, and the sampling data within the overlapping sub-range of the two sub-ranges are the first sampling data of the two adjacent first sub-data blocks.

4. The method according to claim 3, characterized in that The method further comprises: Obtaining an identifier of each sub-computing node, and establishing an association relationship between the identifier and a first sub-data block to be sent to the sub-computing node; The sending of the plurality of first sub-data blocks to the corresponding sub-operation nodes comprises: For each first sub-data block, the first sub-data block is sent to a corresponding sub-operation node according to the association relationship.

5. The method according to claim 2, characterized in that: The step of fusing the first target data in each two adjacent second sub-data blocks based on the respective signal-to-noise ratios of the two second sub-data blocks to the multiple repeated target data comprises: Obtaining a signal-to-noise ratio of each of the two second sub-data blocks; Determining respective weights of the two second sub-data blocks according to the respective signal-to-noise ratios of the two second sub-data blocks; The first target data of each of the two second sub-data blocks are weightedly summed according to the weights to obtain fused data.

6. The method according to any one of claims 1 to 5, characterized in that: The main computing node and the multiple sub-computing nodes are deployed based on an independent node cluster, or the main computing node and the multiple sub-computing nodes are deployed based on a K8s cluster, and the data processing model is loaded on each sub-computing node based on the Kubeflow management platform to process the first sub-data block input therein.

7. A seismic data processing device, characterized in that: A main operation node of a seismic data processing system for parallel processing of seismic data, the seismic data processing system comprising the main operation node and a plurality of sub-operation nodes; the device comprising: An input module, configured to receive first seismic data and business information using the first seismic data; the first seismic data includes a plurality of sampled data, and the business information indicates a target business; A slicing module, configured to slice the first seismic data in time and space to obtain a plurality of first sub-data blocks sequentially arranged in time and space dimensions, wherein each first sub-data block includes a plurality of sample data; A computing module, used for sending the plurality of first sub-data blocks to the corresponding sub-computing nodes respectively, and for the data processing models of the plurality of sub-computing nodes respectively to infer the plurality of sampled data of the first sub-data blocks in parallel, to obtain inference results respectively corresponding to the plurality of first sub-data blocks; A splicing module is used to obtain multiple reasoning results from multiple sub-operation nodes, and splice the multiple reasoning results according to the arrangement order of the first sub-data blocks to obtain a target result.

8. A seismic data processing device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 6 when executed by a processor.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images