A 3D prestack reverse time migration method based on FPGA using wavefield reconstruction

Abstract

The present invention relates to the technical field of oil and gas exploration, and particularly to a three-dimensional prestack reverse time migration method based on FPGA using wavefield reconstruction; this method takes the collected seismic data as the boundary data input of the two-way wave equation, solves the wave equation starting from the shot point to obtain the forward wavefield, and at the same time obtains the reverse wavefield (i.e., the backward wavefield) value by propagating backward in time. The imaging profile is obtained by performing zero-delay cross-correlation (using the cross-correlation imaging condition) operation on the forward wavefield and the backward wavefield at the corresponding time, and finally the imaging result is obtained and the imaging result is output; through this method, the present invention greatly reduces the requirement for data storage, improves the overall system calculation efficiency by trading calculation for space, and meets the calculation requirements of 3D-RTM.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas exploration, and in particular to a three-dimensional prestack reverse time migration method based on FPGA using wave field reconstruction. Background Art

[0002] In recent years, due to the acute shortage of oil and gas resources in China, the state has put forward higher requirements for the development and utilization of energy. To find new oil and gas resources, not only the output of discovered oil and gas reservoirs needs to be maximized, but also new oil and gas fields need to be explored in increasingly complex areas. In the exploration of complex oil and gas reservoirs, seismic imaging is playing an increasingly important role as a tool. Reverse time migration (RTM) has the following advantages over other migration methods: high imaging accuracy, no dip limitation, etc., which can meet the requirements of migration imaging for complex geological structures. Reverse time migration mainly includes three key steps: forward extrapolation of the source wave field based on the two-way wave equation, backward extrapolation of the receiver wave field, and application of the imaging condition. The implementation process can be briefly described as follows: First, the source wave field is forward extrapolated using the two-way wave equation, and the extrapolated wave field is saved; then, the receiver wave field is backward extrapolated using the two-way wave equation. For each backward extrapolation step, the imaging condition is applied for calculation to obtain a local imaging data volume; finally, the reverse time migration results of all shot gathers are stacked to obtain the final prestack depth migration imaging result. Reverse time migration has obvious advantages over other methods in terms of imaging accuracy. However, reverse time migration based on the wave equation is a high-density computational task, requiring huge amounts of computational and storage resources for data processing. In recent years, with the rapid development of computer technology and the significant improvement in computing power, solutions to the large computational and storage requirements of reverse time migration have been studied. Large-scale computational methods and architectures for industrial applications have also been studied.

[0003] For a long time, the improvement of computing performance has mainly benefited from the increase in the operating frequency of microprocessors, and Moore's Law has dominated the pace of the development of information technology. The problem of system heating caused by the increase in the main frequency has become increasingly prominent, and the system power consumption has been rising continuously. These problems are also constantly restricted by physical factors. To ensure the continuous improvement of the performance of microprocessor chips, and more importantly, to reduce the chip power consumption and complexity, the current mainstream commercial CPU designs have comprehensively adopted multi-threaded multi-core architectures. Dual-core and quad-core have become the mainstream products of CPUs, and six-core and eight-core CPU products have also been mass-produced on a large scale. It is expected that the continuous increase in processor cores (referred to as many-core processors) will become the main feature of the development of CPU technology in the future. How to develop multi-core parallel computing programs quickly and effectively is crucial for giving full play to the performance of multi-core processor systems. In addition to the traditional technical route of the development of CPUs from single-core to multi-core (many-core), there are several new technical routes that are currently active, and they may represent important directions for the development of future high-performance computing technologies. An important direction is the reconfigurable computing technology based on FPGA (Field Programmable Gate Array). Whether it is internationally or domestically, the petaflop computer systems representing the current highest computing level have adopted this heterogeneous parallel computing system architecture. How to build a multi-level parallel computing software development framework and programming tools on heterogeneous parallel computing systems to promote the development and transplantation of large-scale parallel computing application software is the key to realizing the large-scale popularization and application of heterogeneous parallel computing systems. The development of heterogeneous cooperative parallel computing technology (FPGA) parallel computing technology can improve the computing level of the large amount of calculations of the entire geophysical data, creating more economic benefits for society; at the same time, various technological advancements can also promote the reduction of the costs of computer software and hardware, introducing more computers to engage in high-performance computing. Summary of the Invention

[0004] In view of the above technical problems, the present invention provides a three-dimensional prestack reverse time migration method based on FPGA using wavefield reconstruction. The present invention can greatly increase the computing efficiency of RTM, reduce the data volume in the entire computing process, and at the same time meet the computing requirements of the single-shot number of current high-density acquisitions and deep and ultra-deep imaging.

[0005] A three-dimensional prestack reverse time migration method based on FPGA using wavefield reconstruction, wherein the method includes:

[0006] Step S1: Input the collected seismic data as the boundary data of the two-way wave equation, and solve the wave equation starting from the shot point to obtain the forward-propagating wavefield;

[0007] Step S2: Store the boundary of the forward-propagating wavefield in the CPU medium;

[0008] Step S3: After the boundaries are stored in the CPU medium, record the reverse wavefield of the data. Then, reconstruct the forward wavefield using the difference equation and read the forward wavefield boundaries on the CPU medium for overwriting.

[0009] Step S4: Perform zero-lag cross-correlation operation on the forward wavefield and the reverse wavefield to obtain the imaging result.

[0010] Step S5: Transmit the cross-correlated imaging result back to the CPU medium, complete data stacking on the CPU medium side, and output the finally processed imaging result.

[0011] In the above method for three-dimensional prestack reverse time migration based on FPGA using wavefield reconstruction, in step S1, prepare the seismic data required for forward continuation on the CPU side, including preparing the constant difference coefficients, velocity parameters, boundary absorption parameters, and source wavelet parameters required during the calculation process.

[0012] In the above method for three-dimensional prestack reverse time migration based on FPGA using wavefield reconstruction, in step S1, perform data interaction with the FPGA through the hardware PCIE interface, transmit the constant difference coefficients, velocity parameters, boundary absorption parameters, and source wavelet parameters to the FPGA for single-shot forward continuation calculation, and store the effective boundaries corresponding to each moment in the CPU medium in step S2.

[0013] In the above method for three-dimensional prestack reverse time migration based on FPGA using wavefield reconstruction, in step S3, adopt the second-order time difference equation to calculate the boundary data of the forward wavefield at the last two moments as the initial values of the wavefield for reverse reconstruction, and cache the initial values of the wavefield at these two moments on the FPGA side; then perform reverse time continuation, and reconstruct the forward wavefield using the recorded initial values of the forward wavefield, and read the effective boundaries from the CPU side for boundary overwriting.

[0014] In the above method for three-dimensional prestack reverse time migration based on FPGA using wavefield reconstruction, in step S4, perform zero-lag cross-correlation operation on the forward wavefield and the reverse wavefield at the same moment to obtain the imaging result of the cross-correlation of single-shot data.

[0015] In the above method for three-dimensional prestack reverse time migration based on FPGA using wavefield reconstruction, in step S5, perform low-frequency filtering processing after completing data stacking on the CPU medium side, and then output the finally processed imaging result.

[0016] The above technical solution has the following advantages or beneficial effects:

[0017] The 3D prestack reverse time migration method based on FPGA using wavefield reconstruction utilizes the high parallel computing advantage of FPGA to design a 3D-RTM (3D prestack reverse time migration) computing system for wavefield reconstruction method implemented by FPGA. The FPGA is used for the forward wavefield calculation once, and the effective boundary data of the wavefield is recorded in the CPU medium. During the reverse continuation process of the gather data, the wavefield values of the forward propagation are reconstructed using the difference equation. In this way, the requirement for data storage is greatly reduced, and the overall computing efficiency of the system is improved by trading computing for space, meeting the computing requirements of 3D-RTM. Brief Description of the Drawings

[0018] By reading the detailed description of the non-restrictive embodiments with reference to the following drawings, the present invention and its features, shape, and advantages will become more obvious. The same reference numerals indicate the same parts in all the drawings. The drawings are not drawn to scale, and the focus is on showing the gist of the present invention.

[0019] Figure 1 It is a flowchart of the 3D prestack reverse time migration method based on FPGA using wavefield reconstruction. Detailed Embodiments

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] The present invention realizes the 3D-RTM algorithm of the wavefield reconstruction method by using the heterogeneous architecture of FPGA and CPU. The large amount of computation of 3D-RTM is implemented in the FPGA, and the forward wavefield calculation data is stored on the CPU platform using the high-speed data interaction interface of PCIE. Utilizing the large parallel computing advantage of FPGA, trading time for space greatly improves the overall computing time of 3D-RTM, and at the same time greatly reduces the consumption of the entire computing space. Through the calculation implemented by wavefield reconstruction in the FPGA, the overall utilization rate of hardware resources is greatly improved, and the requirements for the hardware resources of the entire computing platform are also significantly reduced. During the calculation process in the FPGA, the data is transmitted back to the CPU to balance the computing time and data transmission time.

[0022] The specific process of the 3D prestack reverse time migration method based on FPGA using wavefield reconstruction of the present invention is as Figure 1 shown, and the method includes:

[0023] Step S1: Input the collected seismic data as the boundary data of the two-way wave equation, and solve the wave equation starting from the shot point to obtain the forward propagation wave field;

[0024] Step S2: Store the boundary of the forward propagation wave field in the CPU medium;

[0025] Step S3: After the boundary is stored in the CPU medium, record the reverse wave field of the data. Then, reconstruct the forward propagation wave field using the difference equation and read the boundary of the forward propagation wave field on the CPU medium for overwriting;

[0026] Step S4: Perform zero-delay cross-correlation operation on the forward propagation wave field and the reverse wave field to obtain the imaging result;

[0027] Step S5: Transmit the cross-correlated imaging result back to the CPU medium, complete data stacking on the CPU medium side, and output the finally processed imaging result.

[0028] In the technical solution of the present invention, for 3D-RTM seismic imaging, the collected seismic data is input as the boundary data of the two-way wave equation. Starting from the shot point, the wave equation is solved to obtain the forward propagation wave field. At the same time, the reverse propagation wave field (i.e., the reverse wave field) value is obtained by reverse-time propagation in time. The imaging profile is obtained by performing zero-delay cross-correlation (using the cross-correlation imaging condition) on the forward propagation wave field and the reverse wave field at the corresponding time. For the data acquisition of actual seismic exploration, a mobile observation system is adopted, and the single-shot data for imaging calculation are independent of each other; in the entire calculation design process, the single-shot data is used as the calculation unit, and a complete forward modeling process is completed through one forward modeling calculation. The effective boundary is recorded in the CPU medium. When performing reverse-time continuation of the recorded data wave field, the forward propagation wave field is reconstructed using the finite difference equation and the effective boundary on the CPU is read for overwriting. Finally, the forward propagation wave field and the reverse wave field at the same moment are subjected to cross-correlation imaging, and the imaging result is finally obtained and the imaging result is output.

[0029] In the embodiment of the present invention, the specific implementation process is as follows:

[0030] 1) Prepare the calculation data required for forward modeling (forward continuation) on the CPU side, and prepare the constant difference coefficient, velocity parameter, boundary absorption parameter, source wavelet, observation system and other parameter information required during the calculation process. The programming is completed using the C++ programming language in the Linux operating system environment.

[0031] 2) Perform data interaction with the FPGA through the hardware (PCIE) interface, and transmit parameters such as velocity parameters, source wavelets, constant difference coefficients, velocity parameters, and boundary absorption parameters to the FPGA for single-shot forward continuation calculation, and store the corresponding effective boundary at each moment in the CPU medium.

[0032] 3) Calculate the boundary data of the forward wave field at the last two moments (using a second-order time-difference equation) as the initial value of the wave field for inverse reconstruction. Cache the wave field values at these two moments on the FPGA side or send them back to the CPU side (depending on the hardware resources).

[0033] 4) Extend the gather data backward in time and reconstruct the forward wave field using the recorded initial value of the forward wave field, and read the valid boundary from the CPU side for boundary coverage.

[0034] 5) Calculate the cross-correlation between the forward wave field and the inverse wave field at this moment to obtain the cross-correlation imaging result of the single-shot data.

[0035] 6) Send the cross-correlation result back to the CPU, complete data stacking and subsequent processing on the CPU side, such as performing low-frequency filtering processing, and output the finally processed imaging result.

[0036] The present invention uses an FPGA to perform a forward wave field calculation once, records the effective boundary data of the wave field in the CPU medium, and reconstructs the forward wave field values using a difference equation during the backward extension process of the gather data. In this way, the requirement for data storage is greatly reduced, and the overall system calculation efficiency is improved by trading calculation for space, meeting the calculation requirements of 3D-RTM. Through pressure testing with a large-scale three-dimensional seismic data volume, the results show that the inventive method can meet the calculation requirements of 3D-RTM for ultra-large-scale seismic data, and compared with the same CPU acceleration scheme, its acceleration performance can reach an acceleration ratio of 100 times.

[0037] The algorithm of the wave field reconstruction method implemented by the heterogeneous architecture of FPGA and CPU in the present invention is a 3D-RTM seismic imaging algorithm, which greatly reduces the requirement for data storage, improves the overall system calculation efficiency by trading calculation for space, and meets the calculation requirements of 3D-RTM. By adopting a pipelined data flow method, the data transmission pressure of the effective boundary can be fully met. The patent of the present invention greatly reduces the amount of data required in the calculation process by means of wave field reconstruction, greatly reduces the pressure on the hardware memory requirements, and provides feasible computing power support for deep and ultra-deep formation imaging currently faced by the industry. And the entire implementation of the present invention gives full play to the parallel computing advantages of the FPGA, makes full use of the balance between the computing time and the data transmission time during the entire calculation process, and greatly improves the utilization rate of the hardware computing resources. By this way of submerging the data transmission time in the computing time, the I / O bottleneck problem is eliminated.

[0038] Those skilled in the art should understand that those skilled in the art can implement the said variation examples in combination with the prior art and the above embodiments, which will not be elaborated here. Such variation examples do not affect the essence of the present invention and will not be elaborated here.

[0039] The preferred embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and the devices and structures not described in detail should be understood to be implemented in a common manner in the art; any person skilled in the art can make many possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above, or modify it into an equivalent embodiment with equivalent changes, which does not affect the essence of the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A 3D prestack reverse time migration method based on FPGA using wavefield reconstruction, characterized in that The method includes: Step S1: Input the collected seismic data as the boundary data of the two-way wave equation, and solve the wave equation starting from the shot point to obtain the forward wave field; Step S2: Store the boundary of the forward wave field in the CPU medium; Step S3: After the boundary is stored in the CPU medium, record the reverse wave field of the data, and then reconstruct the forward wave field using the difference equation and read the forward wave field boundary on the CPU medium for overwriting; Step S4: Perform zero-delay cross-correlation operation on the forward wave field and the reverse wave field to obtain the imaging result; Step S5: Transmit the cross-correlated imaging result back to the CPU medium, complete data stacking on the CPU medium side, and output the finally processed imaging result; Among them, the CPU side prepares the seismic data required for forward continuation, including preparing the constant difference coefficient, velocity parameter, boundary absorption parameter, and source wavelet parameter required in the calculation process; In step S1, data interaction is performed with the FPGA through the hardware PCIE interface, and the constant difference coefficient, velocity parameter, boundary absorption parameter, and source wavelet parameter are transmitted to the FPGA for single-shot forward continuation calculation, and the effective boundary corresponding to each moment is stored in the CPU medium in step S2; In step S3, a second-order time difference equation is used to calculate the boundary data of the forward wave field at the last two moments as the initial value of the wave field for reverse reconstruction, and the initial values of the wave fields at these two moments are cached on the FPGA side; then reverse continuation is performed, and the forward wave field is reconstructed using the recorded initial value of the forward wave field, and the effective boundary is read from the CPU side for boundary overwriting.

2. The method for three-dimensional prestack reverse time migration based on FPGA using wavefield reconstruction according to claim 1, wherein In step S4, perform zero-delay cross-correlation operation on the forward wave field and the reverse wave field at the same moment to obtain the imaging result of the cross-correlation of single-shot data.

3. The 3D pre-stack reverse time migration method based on FPGA using wave field reconstruction according to claim 1, wherein In step S5, after data stacking is completed on the CPU medium side, low-frequency filtering processing is performed, and then the finally processed imaging result is output.

Images