Oil and gas exploration method and system based on intelligent geophone

Abstract

This invention provides an oil and gas exploration method and system based on intelligent geophones, relating to the field of artificial intelligence technology. In this invention, geophone data of the target exploration area is extracted; the network parameters of a candidate oil and gas feature mining network are optimized and adjusted to obtain an oil and gas feature mining network; for the geophone data of the exploration area, the oil and gas feature mining network is used to mine the corresponding features of the exploration area to be processed; the features of the exploration area to be processed are compared and analyzed with the corresponding features of each potential oil and gas exploration scheme in a pre-configured set of potential oil and gas exploration schemes to determine the most suitable target number of potential oil and gas exploration schemes; and based on the most suitable target number of potential oil and gas exploration schemes, the target oil and gas exploration scheme for the target exploration area is obtained. Based on the above, the problem of low efficiency in determining existing oil and gas exploration schemes can be improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to an oil and gas exploration method and system based on an intelligent detector. Background Technology

[0002] A geophone is a geophysical device primarily used in seismic exploration. It detects and records underground vibration or sound waves. In oil and gas exploration, geophones are typically used to receive seismic waves generated by a source (such as a vibrating engine or blasting). These waves are reflected after passing through underground rock layers and captured by a geophone array on the surface. The source generates seismic waves, which propagate underground and encounter different strata (such as oil and gas reservoirs), producing reflected waves. These reflected waves return to the surface at different time intervals and with varying intensities. A geophone array on the surface captures these reflected waves. Each geophone converts these vibration signals into electrical signals. The recorded electrical signals are transmitted to a data processing center. By analyzing the intensity, time difference, and waveform of these signals, the structure of the underground rock layers can be inferred, allowing for the generation of oil and gas exploration plans based on the corresponding structure for exploration operations. However, in current technologies, the generation of oil and gas exploration plans requires the participation of relevant personnel throughout the analysis process, resulting in low efficiency. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide an oil and gas exploration method and system based on an intelligent geophone, so as to improve the problem of low efficiency in determining oil and gas exploration schemes in the prior art.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0005] A method for oil and gas exploration based on a smart geophone includes:

[0006] The exploration area detection data of the target exploration area is extracted, and the exploration area detection data belongs to the seismic wave data collected by at least one detector deployed in the target exploration area;

[0007] The network parameters of the candidate oil and gas feature mining network are optimized and adjusted to obtain the oil and gas feature mining network. The optimization and adjustment of the candidate oil and gas feature mining network is based on at least two dimensions of error indicators.

[0008] Based on the detection data of the exploration area, the oil and gas feature mining network is used to mine the corresponding exploration area features to be processed from the detection data of the exploration area.

[0009] The characteristics of the exploration area to be processed are compared and analyzed with the characteristics of each of the pre-configured oil and gas exploration schemes in the set of pending oil and gas exploration schemes to determine the most suitable target number of pending oil and gas exploration schemes. Based on the most suitable target number of pending oil and gas exploration schemes, the target oil and gas exploration scheme for the target exploration area is obtained.

[0010] In some preferred embodiments, in the above-described oil and gas exploration method based on intelligent geophones, the step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network includes:

[0011] The target sample detection data cluster is extracted. The target sample detection data cluster includes target sample detection data and a first number of comparison sample detection data clusters. Each comparison sample detection data cluster includes a related sample detection dataset and an unrelated sample detection dataset. The related sample detection datasets belonging to different comparison sample detection data clusters have different degrees of similarity to the target sample detection data.

[0012] Using a candidate oil and gas feature mining network, the sample exploration area features of each sample exploration area in the target sample detection data cluster are mined out. Each sample exploration area detection data in the target sample detection data cluster corresponds to at least one sample oil and gas exploration scheme.

[0013] Based on the characteristics of the sample exploration area of ​​the detector data of each sample exploration area, the feature comparison analysis network is used to determine the fitness distribution parameters of the detector data of each sample exploration area. The fitness distribution parameters include a second number of exploration scheme fitness parameters, and each exploration scheme fitness parameter corresponds to a sample oil and gas exploration scheme.

[0014] Based on the fit distribution parameters of the detector data of each sample exploration area and at least one sample oil and gas exploration scheme corresponding to the detector data of each sample exploration area, the corresponding scheme fit error parameters are calculated.

[0015] Based on the sample exploration area characteristics of the detection data of each sample exploration area, the corresponding detection comparison error parameter is calculated. The detection comparison error parameter is determined based on a first number of local detection comparison error parameters. Each local detection comparison error parameter is used to reflect the difference between at least one comparison sample detection data cluster and the target sample detection data.

[0016] Based on the scheme adaptation error parameters and the detection comparison error parameters, the network parameters of the candidate oil and gas feature mining network are optimized and adjusted to form a corresponding oil and gas feature mining network.

[0017] This invention also provides an oil and gas exploration system based on an intelligent geophone, including a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program to implement the above-described oil and gas exploration method based on an intelligent geophone.

[0018] The oil and gas exploration method and system based on intelligent geophones provided in this invention can extract geophone data of the target exploration area; optimize and adjust the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network; for the geophone data of the exploration area, use the oil and gas feature mining network to mine the corresponding exploration area features to be processed; compare and analyze the exploration area features to be processed with the corresponding oil and gas exploration scheme features of each of the pre-configured pending oil and gas exploration schemes to determine the most suitable target number of pending oil and gas exploration schemes; and, based on the most suitable target number of pending oil and gas exploration schemes, obtain the target oil and gas exploration scheme for the target exploration area. Based on the foregoing, since the oil and gas feature mining network formed through training can mine the corresponding features of the exploration area to be processed from the detection data of the exploration area, the most suitable number of potential oil and gas exploration schemes can be determined based on the features of the exploration area to be processed. Then, based on this, the target oil and gas exploration scheme can be further determined. For example, the target oil and gas exploration scheme can be selected from the most suitable number of potential oil and gas exploration schemes or modified and adjusted to obtain the target oil and gas exploration scheme in response to manual operation. Compared with the existing technology that directly responds to the scheme generation operation performed by the staff based on the detection data of the exploration area, it can be more efficient. Therefore, it can improve the problem of low efficiency in determining oil and gas exploration schemes in the existing technology.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0020] Figure 1 This is a structural block diagram of an oil and gas exploration system based on a smart detector, provided in an embodiment of the present invention.

[0021] Figure 2 This is a flowchart illustrating the steps of the oil and gas exploration method based on a smart detector provided in this embodiment of the invention.

[0022] Figure 3 This is a schematic diagram of the spectrum of detection data in an exploration area provided in an embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of the modules included in the oil and gas exploration device based on a smart detector provided in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0025] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0026] like Figure 1 As shown in the figure, this embodiment of the invention provides an oil and gas exploration system based on a smart geophone. The oil and gas exploration system based on a smart geophone may include a memory and a processor.

[0027] In detail, the memory and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, they can be electrically connected via one or more communication buses or signal lines. The memory may store at least one software functional module (computer program) that exists in the form of software or firmware. The processor can be used to execute the executable computer program stored in the memory, thereby implementing the oil and gas exploration method based on a smart detector provided in this embodiment of the invention.

[0028] Specifically, in one possible implementation, the memory may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The processor may be a general-purpose processor, including a Central Processing Unit (CPU), Network Processor (NP), System on Chip (SoC), etc.; it may also be a Digital Signal Processor (DSP), Application-Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0029] Specifically, in one possible implementation, the oil and gas exploration system based on the smart detector can be a server with data processing capabilities.

[0030] Combination Figure 2 This invention also provides an oil and gas exploration method based on a smart geophone, which can be applied to the aforementioned oil and gas exploration system based on a smart geophone. The method steps defined in the relevant process of the oil and gas exploration method based on the smart geophone can be implemented by the oil and gas exploration system based on the smart geophone. The following will describe... Figure 2 The specific process shown will be explained in detail.

[0031] Step S110: Extract the exploration area detection data of the target exploration area.

[0032] In this embodiment of the invention, the oil and gas exploration system based on intelligent geophones can extract exploration area geophone data of the target exploration area. The exploration area geophone data is seismic wave data acquired by at least one geophone deployed in the target exploration area; exemplarily, it can be acquired by a geophone array comprising multiple geophones. For example, firstly, a seismic source, such as a vibrating vehicle or an explosion, can be used to generate seismic waves. These waves propagate downwards, passing through various underground rock layers. When the seismic waves encounter different rock layers, reflection and refraction occur. For instance, when a seismic wave enters a shale layer from a sandstone layer, due to the difference in density and propagation speed, part of the wave will be reflected back. Detectors on the ground will capture the reflected waves. Different rock layers will produce reflected waves at different times. The intensity of the reflected waves can reflect information about the reflection interface. For example, a high intensity may indicate that the seismic wave is reflected from low-density rock to high-density rock. By measuring the time it takes for the reflected wave to return to the detector, and based on the time difference and the known seismic wave velocity, the depth of the reflection interface can be calculated. Waveform analysis can provide more information about the underground structure, such as the continuity, thickness, and inclination of the rock layers. In other words, the receiver data from different exploration areas actually represent the characteristic information of different exploration areas. Exploration areas with different characteristic information may require different exploration methods, such as different drilling methods and different equipment, i.e., different oil and gas exploration schemes. Therefore, feature mining of the receiver data from exploration areas can help determine suitable oil and gas exploration schemes. Furthermore, the receiver data from the exploration area can be spectral data. For example, after obtaining seismic wave data, a Fourier transform can be performed to obtain the corresponding spectral data, such as... Figure 3 As shown.

[0033] Step S120: Optimize and adjust the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network.

[0034] In this embodiment of the invention, the oil and gas exploration system based on an intelligent geophone can optimize and adjust the network parameters of a candidate oil and gas feature mining network to obtain an oil and gas feature mining network. The optimization and adjustment of the candidate oil and gas feature mining network is based on at least two dimensions of error indicators. That is, firstly, the error indicators of the sample data in at least two dimensions are analyzed, and then optimization and adjustment can be carried out in the direction of reducing the error indicators, until at least the error indicators converge, such as being less than or equal to a preset error indicator or the decrease in the error indicators being less than or equal to a preset decrease. In addition, the oil and gas feature mining network can be a convolutional neural network (CNN).

[0035] Step S130: For the detection data of the exploration area, the oil and gas feature mining network is used to mine the corresponding exploration area features to be processed from the detection data of the exploration area.

[0036] In this embodiment of the invention, the oil and gas exploration system based on a smart geophone can utilize the oil and gas feature mining network to mine corresponding exploration area features from the geophone data of the exploration area. These exploration area features can reflect the semantic features of the geophone data of the exploration area, that is, characterize the regional semantic features of the exploration area; furthermore, the representation of these exploration area features can be a vector.

[0037] Step S140: Compare and analyze the characteristics of the exploration area to be processed with the characteristics of the corresponding oil and gas exploration schemes of each of the pre-configured oil and gas exploration schemes in the set of pending oil and gas exploration schemes, determine the most suitable target number of pending oil and gas exploration schemes, and obtain the target oil and gas exploration scheme for the target exploration area based on the most suitable target number of pending oil and gas exploration schemes.

[0038] In this embodiment of the invention, the oil and gas exploration system based on a smart geophone can compare and analyze the features of the exploration area to be processed with the corresponding features of each of the pre-configured oil and gas exploration schemes in a set of pending oil and gas exploration schemes (e.g., by encoding the pending oil and gas exploration schemes using a trained text encoder to obtain the features of the proposed oil and gas exploration schemes). This analysis determines the most suitable target number of pending oil and gas exploration schemes, and based on the most suitable target number of pending oil and gas exploration schemes, a target oil and gas exploration scheme for the target exploration area is obtained. Here, "suitability" can refer to the feature similarity (e.g., cosine similarity) between the features of the exploration area to be processed and the features of the oil and gas exploration schemes being greater than or equal to a preset feature similarity, such as 0.5, 0.7, etc. Specifically, the features of the exploration area to be processed can be used as input to a feature comparison analysis network, which outputs the suitability distribution parameters of the features of the exploration area to be processed. The feature comparison analysis network (which can be optimized and adjusted together with the oil and gas feature mining network) can be a fully connected head (FC-head), which may include multiple fully connected layers, for example, two fully connected layers. Specifically, the features of the exploration area to be processed are used as input to the feature comparison analysis network, and the network outputs the fitness distribution parameters of the features of the exploration area to be processed. Each fitness distribution parameter includes a second number of exploration scheme fitness parameters, and each exploration scheme fitness parameter corresponds to a potential oil and gas exploration scheme (i.e., the similarity between the features of the exploration scheme and the corresponding potential oil and gas exploration scheme). For ease of explanation, assume that the fitness distribution parameters of a certain exploration area feature to be processed are (0.2, 0.7, 0.9), where a fitness parameter of 0.2 indicates that the probability of belonging to potential oil and gas exploration scheme A is 0.2, and a fitness parameter of 0.7 indicates that the probability of belonging to potential oil and gas exploration scheme B is 0.7. The exploration scheme fit parameter of 0.9 indicates that the probability of belonging to the pending oil and gas exploration scheme C is 0.9. The target number of pending oil and gas exploration schemes that are most suitable can refer to the target number of pending oil and gas exploration schemes with the highest probability, such as one, two, five, etc.

[0039] Based on the foregoing, since the oil and gas feature mining network formed through training can mine the corresponding features of the exploration area to be processed from the detection data of the exploration area, the most suitable number of potential oil and gas exploration schemes can be determined based on the features of the exploration area to be processed. Then, based on this, the target oil and gas exploration scheme can be further determined. For example, the target oil and gas exploration scheme can be selected from the most suitable number of potential oil and gas exploration schemes or modified and adjusted to obtain the target oil and gas exploration scheme in response to manual operation. Compared with the existing technology that directly responds to the scheme generation operation performed by the staff based on the detection data of the exploration area, it can be more efficient. Therefore, it can improve the problem of low efficiency in determining oil and gas exploration schemes in the existing technology.

[0040] For example, in a simple exemplary oil and gas exploration scheme, a rotary drilling rig can be selected, equipped with multi-functional logging tools to measure porosity, permeability, and reservoir gas content in real time. Specialized carbonate drill bits are used to adapt to hard rock, with drilling depths up to 2000 meters. Geological characteristics: Carbonate rocks, a type of rock including limestone and dolomite, typically formed on the seabed, where oil and gas are stored in the form of pores and fractures.

[0041] For example, in a simple exemplary oil and gas exploration scheme, horizontal directional drilling technology is employed, including horizontal wells and multi-sidehole horizontal wells, using a high-performance hydraulic drilling rig to achieve precise orientation and control, with a drilling depth of 3500 meters. Geological characteristics: Shale, a layered rock where gas is stored in the micropores of the rock, thus typically requiring horizontal wells and high-pressure water fracturing technology to release the gas.

[0042] For example, in a simple exemplary oil and gas exploration scheme, a high-speed hydraulic drilling rig, equipped with a salt rock-specific drill bit, is used to ensure stable salt cavern drilling, conducting internal measurements, geological monitoring, and sealing work within the salt cavern, at a drilling depth of 800 meters. Geological characteristics: Salt rock, a type of rock that can trap gas; salt caverns are typically underground cavities formed by the dissolution of underground salt rock, which can be used for gas storage.

[0043] Encoding text means converting it into a machine-understandable vector representation. A common approach is to use word embeddings, where each word is mapped to a fixed-dimensional vector in a continuous vector space. Here, we simply separate the text into a series of words using spaces, and then find the corresponding word embedding for each word (in reality, these embeddings are usually pre-trained). Below is an example of encoded text:

[0044] Text: "Utilizing horizontal directional drilling technology, including horizontal wells and multi-sidehole horizontal wells, and employing high-performance hydraulic drilling rigs, precise orientation and control are achieved, with drilling depths reaching 3500 meters."

[0045] Example encoded vector (exemplary word embedding): [[0.2,0.5,0.1],[0.6,0.3,0.8],[0.4,0.2,0.7],[0.9,0.2,0.4],[0.5,0.7,0.3],[0.3,0.1,0.6],[0.8,0.4,0.2],[0.7,0.9,0.5],[0.2,0.3,0.8],[0.1,0.7,0.4],[0.6,0.5,0.9],[0.2,0.8,0.3],[0.6,0.5,0.7],[0.9,0.1,0.4],[0.3,0.6,0.8],[0.4,0.7,0.2] ,[0.5,0.2,0.9],[0.7,0.4,0.8],[0.5,0.1,0.3],[0.8,0.9,0.6],[0.7,0.4,0.5],[0.5,0.6,0.2],[0.2,0.8,0.1],[0.3,0.9,0.4],[0.7,0.5,0.6],[0.1,0.3,0.7],[0.9,0.2,0.5],[0.5,0.7,0.3],[0.2,0.6,0.8],[0.3,0.5,0.7],[0.4,0.8,0.2],[0.7,0.9,0.6],[0.6,0.2,0.5]).

[0046] Specifically, in one possible implementation, step S120 described above may include:

[0047] A target sample detection data cluster is extracted. The target sample detection data cluster includes target sample detection data and a first number of comparison sample detection data clusters. Each comparison sample detection data cluster includes a relevant sample detection dataset and an irrelevant sample detection dataset. The similarity between the relevant sample detection datasets belonging to different comparison sample detection data clusters and the target sample detection data varies. That is, firstly, a target sample detection data (a type of seismic wave data) can be obtained. Then, at least two comparison sample detection data clusters are constructed based on the target sample detection data. Each comparison sample detection data cluster includes a relevant sample detection dataset (i.e., a positive sample detection dataset) and an irrelevant sample detection dataset (i.e., a negative sample detection dataset). For example, assuming the first number is equal to 2, the similarity between all relevant sample detection data in one comparison sample detection data cluster and the target sample detection data is greater than the similarity between all relevant sample detection data in another comparison sample detection data cluster and the target sample detection data.

[0048] Using a candidate oil and gas feature mining network, the sample exploration area features of each sample exploration area in the target sample detection data cluster are mined. Each sample exploration area in the target sample detection data cluster corresponds to at least one sample oil and gas exploration scheme. That is, one sample exploration area detection data corresponds to one sample oil and gas exploration scheme or multiple sample oil and gas exploration schemes, meaning that multiple sample oil and gas exploration schemes are suitable for exploration.

[0049] Based on the sample exploration area characteristics of the detector data of each sample exploration area, a feature comparison analysis network is used to determine the fitness distribution parameters of the detector data of each sample exploration area. The fitness distribution parameters include a second number of exploration scheme fitness parameters, and each exploration scheme fitness parameter corresponds to a sample oil and gas exploration scheme (as mentioned above, i.e., it is compared and analyzed with each of the pending oil and gas exploration schemes in the set of pending oil and gas exploration schemes).

[0050] Based on the fit distribution parameters of the detector data for each of the sample exploration areas (i.e., the predicted fit) and at least one sample oil and gas exploration scheme corresponding to the detector data for each of the sample exploration areas (i.e., the actual fit), the corresponding scheme fit error parameter is calculated to characterize the difference between the actual fit and the predicted fit. Specifically, this can be achieved using the cross-entropy function ([H(p,q)=-\sum_{x}p(x)\log q(x)]; where (x) represents all possible events, (p(x)) is the actual probability distribution, and (q(x)) is the predicted probability distribution) calculates the adaptation error parameter of the scheme; for example, the various undetermined oil and gas exploration schemes in the set of undetermined oil and gas exploration schemes are undetermined oil and gas exploration scheme A, undetermined oil and gas exploration scheme B and undetermined oil and gas exploration scheme C, and a sample exploration area detection data corresponds to a sample oil and gas exploration scheme. For example, the actual sample oil and gas exploration scheme of sample exploration area detection data 1 is C. Thus, the actual adaptation situation can be (0, 0, 1), the predicted adaptation situation can be (0.3, 0.5, 0.8), and the cross-entropy function error can be 0.22314;

[0051] Based on the sample exploration area characteristics of the detection data of each sample exploration area, a corresponding detection contrast error parameter is calculated. The detection contrast error parameter is determined based on a first number of local detection contrast error parameters. Each local detection contrast error parameter is used to reflect the difference between at least one cluster of contrast sample detection data and the target sample detection data. In this way, the detection contrast error parameter can be used to reduce the difference between relevant sample detection data and target sample detection data, and increase the difference between unrelated sample detection data and target sample detection data.

[0052] Based on the scheme adaptation error parameters and the detection contrast error parameters, the network parameters of the candidate oil and gas feature mining network are optimized and adjusted to form a corresponding oil and gas feature mining network; for example, the network parameters of the candidate oil and gas feature mining network can be optimized and adjusted along the direction of reducing the scheme adaptation error parameters and the detection contrast error parameters until convergence.

[0053] Specifically, in one possible implementation, the first number of comparative sample detection data clusters may include a first type of data cluster and a second type of data cluster. Based on this, the step of extracting the target sample detection data cluster may include:

[0054] In a pre-defined oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-defined oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data. For example, the original sample detection data can be obtained by collecting data from the exploration area using a geophone, and the initial exploration area features can be mined based on other convolutional neural networks (which can be existing convolutional neural networks trained by other mining tasks). As long as feature mining can be performed on the detection data, there is no specific limitation.

[0055] Based on the initial exploration area characteristics, in the preset oil and gas exploration database, the first type of data cluster is determined. The order of the similarity between the relevant sample detection data included in the first type of data cluster and the target sample detection data is less than or equal to a pre-configured reference order. That is, a specified number of sample detection data with the highest data similarity, such as 40, are determined as relevant sample detection data in the first type of data cluster. The unrelated sample detection data included in the first type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the first type of data cluster (which can be one or more randomly determined; considering that it may not be possible to calculate all relevant sample detection data, in order to avoid generating false negative samples, which may be positive samples, the data similarity between all original sample detection data in a batch of the preset oil and gas exploration database can also be calculated, and false negative samples can be eliminated according to the magnitude of the data similarity). In addition, the target sample detection data can also be used as a relevant sample detection data in the first type of data cluster.

[0056] In the preset oil and gas exploration database, a second type of data cluster is identified. The relevant sample detection data included in the second type of data cluster shares the same high-frequency sample oil and gas exploration scheme as the target sample detection data (that is, the relevant sample detection data and the target sample detection data not only share the same sample oil and gas exploration scheme, but the sample oil and gas exploration scheme also belongs to a high-frequency sample oil and gas exploration scheme). The unrelated sample detection data included in the second type of data cluster belongs to at least one other original sample detection data (which can be one or more randomly determined) outside of the relevant sample detection datasets in the first type of data cluster and the relevant sample detection datasets in the second type of data cluster. The high-frequency sample oil and gas exploration scheme is based on the preset oil and gas exploration data. For each original sample detection data in the database, the number of occurrences of the sample oil and gas exploration scheme is calculated to be greater than a pre-configured reference first number. The reference first number can be 20, 50, etc., and can be determined based on the number of original sample detection data in the preset oil and gas exploration database. That is, if sample oil and gas exploration scheme A has 50 original sample detection data in the preset oil and gas exploration database, it can be considered that sample oil and gas exploration scheme A belongs to high-frequency sample oil and gas exploration schemes; if sample oil and gas exploration scheme B has 5 original sample detection data in the preset oil and gas exploration database, it can be considered that sample oil and gas exploration scheme B does not belong to high-frequency sample oil and gas exploration schemes, but belongs to low-frequency sample oil and gas exploration schemes, as described in the following related descriptions.

[0057] Based on this, by combining the existing initial exploration area features and sample oil and gas exploration schemes of the original sample detection data in the pre-set oil and gas exploration database, multi-granularity similarity data clusters (i.e., the first type of data cluster and the third type of data cluster) were determined. This allows the semantic representation ability of the network to be effectively improved by making full use of existing supervised data on the basis of unsupervised comparative learning.

[0058] Specifically, in one possible implementation, the first number of comparative sample detection data clusters may include a first type of data cluster and a third type of data cluster. Based on this, the step of extracting the target sample detection data cluster may include:

[0059] In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data.

[0060] Based on the initial exploration area characteristics, in the preset oil and gas exploration database, the first type of data cluster is determined. The order number of the data similarity (similarity between initial exploration area characteristics) between the relevant sample detection data included in the first type of data cluster and the target sample detection data is less than or equal to the pre-configured reference order number. The unrelated sample detection data included in the first type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the first type of data cluster, as described above.

[0061] In the preset oil and gas exploration database, the third type of data cluster is determined. The relevant sample detection data included in the third type of data cluster have the same oil and gas exploration scheme distribution data as the target sample detection data. The unrelated sample detection data included in the third type of data cluster belongs to at least one other original sample detection data besides the relevant sample detection dataset in the first type of data cluster and the relevant sample detection dataset in the third type of data cluster. The oil and gas exploration scheme distribution data refers to the sample detection data corresponding to each sample oil and gas exploration scheme in each preset oil and gas exploration database (when it is necessary to compare whether two sample detection data have the same oil and gas exploration scheme distribution data, the two sample detection data need to be excluded from the preset oil and gas exploration database). For example, there are sample oil and gas exploration scheme A, sample oil and gas exploration scheme B, and sample oil and gas exploration scheme C. Sample oil and gas exploration scheme A corresponds to sample detection data 1, sample detection data 2, sample detection data 4, and sample detection data 6. Sample oil and gas exploration scheme B corresponds to sample detection data 6. Detection data 2, sample detection data 4, and sample detection data 6 are provided. Sample oil and gas exploration scheme C corresponds to sample detection data 3 and sample detection data 5. At this point, it is necessary to compare sample detection data 1 and sample detection data 2 to determine if the distribution data of the oil and gas exploration schemes are the same. We can first determine the sample oil and gas exploration scheme A corresponding to sample detection data 1, and the sample oil and gas exploration schemes A and B corresponding to sample detection data 2. After removing sample detection data 1 and sample detection data 2, the preset oil and gas exploration database also includes... Given sample detection data 3, sample detection data 4, and sample detection data 5, the oil and gas exploration scheme distribution data for sample detection data 1 is "sample oil and gas exploration scheme A: sample detection data 4, sample detection data 6", and the oil and gas exploration scheme distribution data for sample detection data 2 is "sample oil and gas exploration scheme A: sample detection data 4, sample detection data 6, sample oil and gas exploration scheme B: sample detection data 4, sample detection data 6", meaning they have the same oil and gas exploration scheme distribution data "sample detection data 4, sample detection data 6".

[0062] Based on this, by combining the existing initial exploration area features and sample oil and gas exploration schemes of the original sample detection data in the pre-set oil and gas exploration database, multi-granularity similarity data clusters (i.e., the first type of data cluster and the third type of data cluster) were determined. This allows the semantic representation ability of the network to be effectively improved by making full use of existing supervised data on the basis of unsupervised comparative learning.

[0063] Specifically, in one possible implementation, the first number of comparative sample detection data clusters may include a second type of data cluster and a fourth type of data cluster. Based on this, the step of extracting the target sample detection data cluster may include:

[0064] In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data.

[0065] Based on the initial exploration area characteristics, a fourth data cluster is determined in the preset oil and gas exploration database. The data similarity between the relevant sample detection data included in the fourth data cluster and the initial exploration area characteristics is greater than or equal to the pre-configured reference data similarity. Furthermore, these clusters share the same low-frequency sample oil and gas exploration scheme or have at least two identical sample oil and gas exploration schemes (i.e., each cluster has two or more sample oil and gas exploration schemes, and at least two of these schemes are identical and overlapping, such as the relevant sample detection data having sample oil and gas exploration scheme A, sample oil and gas exploration scheme B, and sample oil and gas exploration scheme C). Gas exploration scheme C, the target sample detection data includes sample oil and gas exploration scheme A, sample oil and gas exploration scheme B and sample oil and gas exploration scheme D), the unrelated sample detection data included in the fourth type of data cluster belongs to at least one other original sample detection data besides the related sample detection dataset in the fourth type of data cluster, the low frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on each original sample detection data in the preset oil and gas exploration database, the occurrence frequency of which is less than the second reference number, the second reference number is less than the first reference number, such as 8 or 10;

[0066] In the preset oil and gas exploration database, the second type of data cluster is extracted. The relevant sample detection data included in the second type of data cluster has the same high-frequency sample oil and gas exploration scheme as the target sample detection data. The unrelated sample detection data included in the second type of data cluster belongs to at least one other original sample detection data besides the relevant sample detection dataset in the fourth type of data cluster and the relevant sample detection dataset in the second type of data cluster. The high-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on the sample oil and gas exploration scheme of each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is greater than the reference first number (as described above).

[0067] Similarly, by combining the existing initial exploration area features and sample oil and gas exploration schemes of the original sample detection data in the pre-set oil and gas exploration database, multi-granularity similarity data clusters (i.e., the second type of data cluster and the fourth type of data cluster) were determined, which enabled the semantic representation ability of the network to be effectively improved by making full use of existing supervised data on the basis of unsupervised comparative learning.

[0068] Specifically, in one possible implementation, the first number of comparative sample detection data clusters may include a third type of data cluster and a fourth type of data cluster. Based on this, the step of extracting the target sample detection data cluster may include:

[0069] In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data.

[0070] Based on the initial exploration area characteristics, the fourth type of data cluster is determined in the preset oil and gas exploration database. The data similarity between the relevant sample detection data included in the fourth type of data cluster and the initial exploration area characteristics is greater than or equal to the pre-configured reference data similarity, and they have the same low-frequency sample oil and gas exploration scheme or at least two identical sample oil and gas exploration schemes. The unrelated sample detection data included in the fourth type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the fourth type of data cluster. The low-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is less than the second reference number, as described above.

[0071] In the preset oil and gas exploration database, the third type of data cluster is extracted. The relevant sample detection data included in the third type of data cluster has the same oil and gas exploration scheme distribution data as the target sample detection data. The unrelated sample detection data included in the third type of data cluster belongs to at least one other original sample detection data other than the relevant sample detection dataset in the fourth type of data cluster and the relevant sample detection dataset in the third type of data cluster.

[0072] Similarly, by combining the existing initial exploration area features and sample oil and gas exploration schemes of the original sample detection data in the pre-set oil and gas exploration database, multi-granularity similarity data clusters (i.e., the third type of data cluster and the fourth type of data cluster) were determined, so that the semantic representation ability of the network can be effectively improved by making full use of the existing supervised data on the basis of unsupervised comparative learning.

[0073] Specifically, in one possible implementation, the first number of comparative sample detection data clusters may include a first type of data cluster, a second type of data cluster, a third type of data cluster, a fourth type of data cluster, and a fifth type of data cluster. Based on this, the step of extracting the target sample detection data cluster may include:

[0074] In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data.

[0075] Based on the initial exploration area characteristics, in the preset oil and gas exploration database, the first type of data cluster is determined. The order of the similarity between the relevant sample detection data and the target sample detection data included in the first type of data cluster is less than or equal to the pre-configured reference order. The unrelated sample detection data included in the first type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the first type of data cluster, as described above.

[0076] Based on the initial exploration area characteristics, the fourth type of data cluster is determined in the preset oil and gas exploration database. The data similarity between the relevant sample detection data included in the fourth type of data cluster and the initial exploration area characteristics is greater than or equal to the pre-configured reference data similarity, and they have the same low-frequency sample oil and gas exploration scheme or at least two identical sample oil and gas exploration schemes (as described above). The unrelated sample detection data included in the fourth type of data cluster belongs to at least one other original sample detection data other than the relevant sample detection dataset in the first type of data cluster and the relevant sample detection dataset in the fourth type of data cluster (the set composed of relevant sample detection data). The low-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is less than the second reference number.

[0077] In the preset oil and gas exploration database, the second type of data cluster is determined. The relevant sample detection data included in the second type of data cluster has the same high-frequency sample oil and gas exploration scheme as the target sample detection data. The unrelated sample detection data included in the second type of data cluster belongs to at least one other original sample detection data besides the relevant sample detection dataset in the first type of data cluster, the relevant sample detection dataset in the fourth type of data cluster, and the relevant sample detection dataset in the second type of data cluster. The high-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on the sample oil and gas exploration scheme of each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is greater than the reference first number. The reference first number is greater than the reference second number (as described above).

[0078] In the preset oil and gas exploration database, the third type of data cluster is determined. The relevant sample detection data included in the third type of data cluster has the same oil and gas exploration scheme distribution data as the target sample detection data. The unrelated sample detection data included in the third type of data cluster belongs to at least one original sample detection data other than the relevant sample detection dataset in the first type of data cluster, the relevant sample detection dataset in the fourth type of data cluster, the relevant sample detection dataset in the second type of data cluster, and the relevant sample detection dataset in the third type of data cluster (as described above).

[0079] In the preset oil and gas exploration database, the fifth type of data cluster is determined. The relevant sample detection data included in the fifth type of data cluster have similar oil and gas exploration scheme distribution data to the target sample detection data. The unrelated sample detection data included in the fifth type of data cluster belongs to at least one other original sample detection data besides the relevant sample detection dataset in the first type of data cluster, the relevant sample detection dataset in the fourth type of data cluster, the relevant sample detection dataset in the second type of data cluster, the relevant sample detection dataset in the third type of data cluster, and the relevant sample detection dataset in the fifth type of data cluster. Similar oil and gas exploration scheme distribution data can refer to the degree of overlap between the sample oil and gas exploration schemes corresponding to the two oil and gas exploration scheme distribution data being greater than or equal to a preset degree of overlap, such as greater than or equal to 0.5, 0.6, 0.8, etc.

[0080] Similarly, by combining the existing initial exploration area features and sample oil and gas exploration schemes of the original sample detection data in the pre-set oil and gas exploration database, multi-granularity similarity data clusters (i.e., the first type of data cluster, the second type of data cluster, the third type of data cluster, the fourth type of data cluster, and the fifth type of data cluster) are determined, so that the semantic representation ability of the network can be effectively improved by making full use of the existing supervised data on the basis of unsupervised comparative learning. Specifically, the data similarity between the relevant sample detection dataset and the target sample detection data in the first data cluster is greater than that between the relevant sample detection dataset and the target sample detection data in the fourth data cluster. The data similarity between the relevant sample detection dataset and the target sample detection data in the fourth data cluster is greater than that between the relevant sample detection dataset and the target sample detection data in the second data cluster. The data similarity between the relevant sample detection dataset and the target sample detection data in the second data cluster is greater than that between the relevant sample detection dataset and the target sample detection data in the third data cluster. The data similarity between the relevant sample detection dataset and the target sample detection data in the third data cluster is greater than that between the relevant sample detection dataset and the target sample detection data in the fifth data cluster.

[0081] Specifically, in one possible implementation, the step of calculating the corresponding scheme adaptation error parameter based on the fit distribution parameter of the detector data for each of the sample exploration areas and at least one sample oil and gas exploration scheme corresponding to the detector data for each of the sample exploration areas may include:

[0082] For each undetermined oil and gas exploration scheme in the second set of sample oil and gas exploration schemes, based on the fit distribution parameters of the detector data of each sample exploration area, the corresponding exploration scheme fit parameters of the undetermined oil and gas exploration scheme are obtained; for each undetermined oil and gas exploration scheme in the second set of sample oil and gas exploration schemes, based on at least one sample oil and gas exploration scheme corresponding to the detector data of each sample exploration area, the corresponding actual fit parameters of the undetermined oil and gas exploration scheme are obtained; for each undetermined oil and gas exploration scheme in the second set of sample oil and gas exploration schemes, based on the exploration scheme fit parameters and actual fit parameters of the detector data of each sample exploration area relative to the undetermined oil and gas exploration scheme, the local scheme fit error parameters corresponding to the undetermined oil and gas exploration scheme are calculated; based on the local scheme fit error parameters corresponding to each undetermined oil and gas exploration scheme, the corresponding scheme fit error parameters are calculated.

[0083] For example, given a sample oil and gas exploration scheme A, sample exploration area detector data 1 and sample exploration area detector data 2, the fitness distribution parameter (i.e., predicted distribution) of sample exploration area detector data 1 is (sample oil and gas exploration scheme A: 0.7), the actual distribution of sample exploration area detector data 1 is (sample oil and gas exploration scheme A: 1), and the fitness distribution parameter (i.e., predicted distribution) of sample exploration area detector data 2 is (sample oil and gas exploration scheme A: 0.2). The actual distribution is (sample oil and gas exploration scheme A: 0); based on this, the exploration scheme adaptation parameters of sample oil and gas exploration scheme A are (sample exploration area detector data 1: 0.7, sample exploration area detector data 2: 0.2), and the actual adaptation parameters of sample oil and gas exploration scheme A are (sample exploration area detector data 1: 1, sample exploration area detector data 2: 0); then, the corresponding local scheme adaptation error parameters (such as the mean values ​​of 0.3567 and 0.2231) can be calculated based on the following formula:

[0084] [L(y,p)=-[y\log(p)+(1-y)\log(1-p)]];

[0085] Where: (L(y,p)) is the error function, representing the difference between the network's prediction and the actual value. (y) is the actual value, which can be 0 or 1. (p) is the network's prediction, and (\log) represents the natural logarithm.

[0086] Specifically, in one possible implementation, the step of calculating the corresponding detection contrast error parameter based on the sample exploration area characteristics of the detector data for each of the sample exploration areas may include:

[0087] Based on the first number of contrast sample detection data clusters and the target sample detection data, a first number of first training detection events and a first number of second training detection events are formed. Each first training detection event includes a relevant sample detection dataset of the contrast sample detection data cluster, and each second training detection event includes a relevant sample detection dataset and / or an unrelated sample detection dataset (i.e., at least one of the two) of the contrast sample detection data cluster. Based on the first number of first training detection events and the first number of second training detection events, a first number of local detection contrast error parameters are calculated. Based on the first number of local detection contrast error parameters, the corresponding detection contrast error parameters are calculated (e.g., by mean calculation).

[0088] To narrow the distance between the target sample detection data and the relevant sample detection data in the comparison sample detection data cluster, and to widen the distance between the target sample detection data and the irrelevant sample detection data in the comparison sample detection data cluster, this application determines a first number of data similarity levels. For each data similarity level, the first training detection event and the second training detection event are calculated to obtain the corresponding local detection contrast error parameter. For example, suppose the first number of comparison sample detection data clusters include a first type of data cluster, a fourth type of data cluster, a second type of data cluster, a third type of data cluster, and a fifth type of data cluster. Therefore, a first number of first training detection events and a first number of second training detection events can be constructed, with a corresponding hierarchical relationship between the first training detection events and the second training detection events. The first training detection event includes at least one first relevant sample detection data pair, and the second training detection event includes at least one first irrelevant sample detection data pair. For ease of explanation, the construction method of the first relevant sample detection data pair and the first irrelevant sample detection data pair corresponding to each data similarity level will be described below.

[0089] (1) The first data similarity level; the first relevant sample detection data pair consists of the target sample detection data and the relevant sample detection data in the first type of data cluster; the first unrelated sample detection data pair consists of the target sample detection data and the unrelated sample detection data in the first type of data cluster.

[0090] (2) The second data similarity level; the first relevant sample detection data pair consists of the target sample detection data and the relevant sample detection data in the first data cluster; the first unrelated sample detection data pair consists of the target sample detection data and the unrelated sample detection data in the fourth data cluster.

[0091] (3) The third data similarity level; the first relevant sample detection data pair consists of the target sample detection data and the relevant sample detection data in the second type of data cluster; the first unrelated sample detection data pair consists of the target sample detection data and any one of the relevant sample detection data or unrelated sample detection data in the third type of data cluster.

[0092] (4) The fourth data similarity level; the first relevant sample detection data pair consists of the target sample detection data and the relevant sample detection data in the fourth data cluster; the first unrelated sample detection data pair consists of the target sample detection data and the unrelated sample detection data in the third data cluster.

[0093] (5) The fifth data similarity level; the first relevant sample detection data pair consists of the target sample detection data and the relevant sample detection data in the second type of data cluster; the first unrelated sample detection data pair consists of the target sample detection data and the unrelated sample detection data in the fifth type of data cluster.

[0094] Based on this, the local detection contrast error parameter corresponding to a data similarity level can be calculated as follows: Calculate the dot product between the target sample detection data and the sample exploration area features corresponding to the relevant sample detection data in the first relevant sample detection data pair, and perform exponential calculation on this dot product to obtain a first exponential value; calculate the dot product between the target sample detection data and the sample exploration area features corresponding to the unrelated sample detection data in the first unrelated sample detection data pair, and perform exponential calculation on this dot product to obtain a second exponential value; take the logarithm of the ratio between the first exponential value and the second exponential value, and finally, invert the result of the logarithm operation to obtain the corresponding local detection contrast error parameter.

[0095] Specifically, in one possible implementation, the step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network may further include:

[0096] At least two target sample detection data sets and a first number of comparison sample detection data clusters corresponding to each target sample detection data set are identified. Based on the at least two target sample detection data sets and the first number of comparison sample detection data clusters corresponding to each target sample detection data set, a third training detection event and a fourth training detection event are formed. The data similarity level corresponding to the relevant sample detection data combination in the third training detection event is greater than the data similarity level corresponding to the unrelated sample detection data combination in the fourth training detection event. Based on the third training detection event and the fourth training detection event, a training comparison error parameter is calculated. As mentioned above, the calculation process for the local detection comparison error parameter for the first and second training detection events can be referred to above. Alternatively, it can be calculated in other ways, such as calculating the feature distance (e.g., cosine distance) between the sample exploration area features corresponding to the sample detection data included in the relevant sample detection data combination in the third training detection event, and calculating the fourth training event. The non-relevant sample detection data combination in the detection event includes the feature distance between the sample exploration area features corresponding to the sample detection data. The relevant sample detection data combination and the non-relevant sample detection data combination are relative. As long as the data similarity level of the former is greater than that of the latter, the data similarity level decreases sequentially from the first data similarity level to the fifth data similarity level. For example, the relevant sample detection data combination may include target sample detection data 1 and the relevant sample detection data in the corresponding first type of data cluster. The non-relevant sample detection data combination may include target sample detection data 2 and the relevant sample detection data in the corresponding second type of data cluster. Target sample detection data 1 and target sample detection data 2 belong to two target sample detection data among the at least two target sample detection data. The difference between the latter feature distance and the former feature distance is calculated, and then the sum of the difference and the target parameter is calculated. Finally, the larger value between the sum and zero is taken as the corresponding training comparison error parameter.The target parameter can be pre-configured or calculated. For example, it can be calculated as the difference between the data similarity level of the relevant sample detection data combination in the third training detection event and the data similarity level of the unrelated sample detection data combination in the fourth training detection event (e.g., for the aforementioned five data similarity levels, the difference between the first and second data similarity levels is 1, the difference between the first and third data similarity levels is 2, the difference between the second and fifth data similarity levels is 3, and so on, thus determining the corresponding difference). Then, the square of this difference can be calculated, and the corresponding training contrast error parameter can be determined based on this square. For example, the training contrast error parameter is positively correlated with this square. Specifically, a configured weight parameter can be multiplied by this square to obtain the training contrast error parameter. This weight parameter can be 0.01, 0.02, 0.03, etc.

[0097] The step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network based on the scheme adaptation error parameters and the detection comparison error parameters to form a corresponding oil and gas feature mining network includes:

[0098] Based on the scheme adaptation error parameters, the detection contrast error parameters, and the training contrast error parameters (such as by first superimposing or other methods of fusion), the network parameters of the candidate oil and gas feature mining network are optimized and adjusted to form a corresponding oil and gas feature mining network.

[0099] Specifically, in one possible implementation, the step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network based on the scheme adaptation error parameters, the detection contrast error parameters, and the training contrast error parameters to form a corresponding oil and gas feature mining network may include:

[0100] The scheme adaptation error parameter, the detection contrast error parameter, and the training contrast error parameter are weighted and summed to calculate the corresponding total error parameter. The specific weight coefficients can be pre-configured, such as 0.5, 0.4, and 0.1 respectively.

[0101] Based on the total error parameter, the network parameters of the candidate oil and gas feature mining network are optimized and adjusted to form a corresponding oil and gas feature mining network. That is, optimization and adjustment are carried out in the direction of reducing the total error parameter so that the total error parameter converges.

[0102] Combination Figure 4This invention also provides an oil and gas exploration device based on a smart geophone, which can be applied to the aforementioned oil and gas exploration system based on a smart geophone. The oil and gas exploration device based on a smart geophone may include:

[0103] The detection data extraction module is used to extract the detection data of the target exploration area, wherein the detection data of the exploration area belongs to the seismic wave data collected by at least one detector deployed in the target exploration area;

[0104] The network optimization module is used to optimize and adjust the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network. The optimization and adjustment of the candidate oil and gas feature mining network is based on at least two dimensions of error indicators.

[0105] The regional feature mining module is used to mine the corresponding exploration area features to be processed from the detection data of the exploration area using the oil and gas feature mining network.

[0106] The oil and gas exploration scheme determination module is used to compare and analyze the characteristics of the exploration area to be processed with the corresponding oil and gas exploration scheme characteristics of each of the pre-configured oil and gas exploration schemes in the set of pending oil and gas exploration schemes, determine the most suitable target number of pending oil and gas exploration schemes, and obtain the target oil and gas exploration scheme for the target exploration area based on the most suitable target number of pending oil and gas exploration schemes.

[0107] In summary, the oil and gas exploration method and system based on intelligent geophones provided by this invention can extract geophone data of the target exploration area; optimize and adjust the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network; for the geophone data of the exploration area, use the oil and gas feature mining network to mine the corresponding exploration area features to be processed; compare and analyze the exploration area features to be processed with the corresponding oil and gas exploration scheme features of each of the pre-configured pending oil and gas exploration schemes to determine the most suitable target number of pending oil and gas exploration schemes; and, based on the most suitable target number of pending oil and gas exploration schemes, obtain the target oil and gas exploration scheme for the target exploration area. Based on the foregoing, since the oil and gas feature mining network formed through training can mine the corresponding features of the exploration area to be processed from the detection data of the exploration area, the most suitable number of potential oil and gas exploration schemes can be determined based on the features of the exploration area to be processed. Then, based on this, the target oil and gas exploration scheme can be further determined. For example, the target oil and gas exploration scheme can be selected from the most suitable number of potential oil and gas exploration schemes or modified and adjusted to obtain the target oil and gas exploration scheme in response to manual operation. Compared with the existing technology that directly responds to the scheme generation operation based on the detection data of the exploration area, it can be more efficient. Therefore, it can improve the problem of low efficiency in determining oil and gas exploration schemes in the existing technology, and can also reduce labor costs to a certain extent.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for oil and gas exploration based on smart geophones, characterized in that, include: The exploration area detection data of the target exploration area is extracted, and the exploration area detection data belongs to the seismic wave data collected by at least one detector deployed in the target exploration area; The network parameters of the candidate oil and gas feature mining network are optimized and adjusted to obtain the oil and gas feature mining network. The optimization and adjustment of the candidate oil and gas feature mining network is based on at least two dimensions of error indicators. Based on the detection data of the exploration area, the oil and gas feature mining network is used to mine the corresponding exploration area features to be processed from the detection data of the exploration area. The characteristics of the exploration area to be processed are compared and analyzed with the characteristics of each of the pre-configured oil and gas exploration schemes in the set of pending oil and gas exploration schemes to determine the most suitable target number of pending oil and gas exploration schemes. Based on the most suitable target number of pending oil and gas exploration schemes, the target oil and gas exploration scheme for the target exploration area is obtained. The candidate oil and gas feature mining network is specifically a convolutional neural network; The step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network includes: The target sample detection data cluster is extracted. The target sample detection data cluster includes target sample detection data and a first number of comparison sample detection data clusters. Each comparison sample detection data cluster includes a related sample detection dataset and an unrelated sample detection dataset. The related sample detection datasets belonging to different comparison sample detection data clusters have different degrees of similarity to the target sample detection data. Using a candidate oil and gas feature mining network, the sample exploration area features of each sample exploration area in the target sample detection data cluster are mined out. Each sample exploration area detection data in the target sample detection data cluster corresponds to at least one sample oil and gas exploration scheme. Based on the characteristics of the sample exploration area of ​​the detector data of each sample exploration area, the feature comparison analysis network is used to determine the fitness distribution parameters of the detector data of each sample exploration area. The fitness distribution parameters include a second number of exploration scheme fitness parameters, and each exploration scheme fitness parameter corresponds to a sample oil and gas exploration scheme. Based on the fit distribution parameters of the detector data of each sample exploration area and at least one sample oil and gas exploration scheme corresponding to the detector data of each sample exploration area, the corresponding scheme fit error parameters are calculated. Based on the sample exploration area characteristics of the detection data of each sample exploration area, the corresponding detection comparison error parameter is calculated. The detection comparison error parameter is determined based on a first number of local detection comparison error parameters. Each local detection comparison error parameter is used to reflect the difference between at least one comparison sample detection data cluster and the target sample detection data. Based on the scheme adaptation error parameters and the detection comparison error parameters, the network parameters of the candidate oil and gas feature mining network are optimized and adjusted to form a corresponding oil and gas feature mining network.

2. The oil and gas exploration method based on a smart detector as described in claim 1, characterized in that, The first number of comparative sample detection data clusters include a first type of data cluster and a second type of data cluster; The step of extracting the target sample detection data cluster includes: In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data. Based on the initial exploration area characteristics, the first type of data cluster is determined in the preset oil and gas exploration database. The order of the similarity between the relevant sample detection data and the target sample detection data included in the first type of data cluster is less than or equal to the pre-configured reference order. The unrelated sample detection data included in the first type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the first type of data cluster. In the preset oil and gas exploration database, the second type of data cluster is determined. The relevant sample detection data included in the second type of data cluster have the same high-frequency sample oil and gas exploration scheme as the target sample detection data. The unrelated sample detection data included in the second type of data cluster belongs to at least one other original sample detection data other than the relevant sample detection dataset in the first type of data cluster and the relevant sample detection dataset in the second type of data cluster. The high-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on the sample oil and gas exploration scheme of each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is greater than the pre-configured reference first number.

3. The oil and gas exploration method based on a smart detector as described in claim 1, characterized in that, The first number of comparative sample detection data clusters include a first type of data cluster and a third type of data cluster; The step of extracting the target sample detection data cluster includes: In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data. Based on the initial exploration area characteristics, the first type of data cluster is determined in the preset oil and gas exploration database. The order of the similarity between the relevant sample detection data and the target sample detection data included in the first type of data cluster is less than or equal to the pre-configured reference order. The unrelated sample detection data included in the first type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the first type of data cluster. In the preset oil and gas exploration database, the third type of data cluster is determined. The relevant sample detection data included in the third type of data cluster has the same oil and gas exploration scheme distribution data as the target sample detection data. The unrelated sample detection data included in the third type of data cluster belongs to at least one other original sample detection data other than the relevant sample detection dataset in the first type of data cluster and the relevant sample detection dataset in the third type of data cluster. The oil and gas exploration scheme distribution data refers to the sample detection data corresponding to each sample oil and gas exploration scheme in each preset oil and gas exploration database.

4. The oil and gas exploration method based on a smart detector as described in claim 1, characterized in that, The first number of comparative sample detection data clusters include a second type of data cluster and a fourth type of data cluster; The step of extracting the target sample detection data cluster includes: In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data. Based on the initial exploration area characteristics, the fourth type of data cluster is determined in the preset oil and gas exploration database. The data similarity between the relevant sample detection data included in the fourth type of data cluster and the initial exploration area characteristics is greater than or equal to the pre-configured reference data similarity, and they have the same low-frequency sample oil and gas exploration scheme or at least two identical sample oil and gas exploration schemes. The unrelated sample detection data included in the fourth type of data cluster belongs to at least one other original sample detection data besides the relevant sample detection dataset in the fourth type of data cluster. The low-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is less than the second reference number. In the preset oil and gas exploration database, the second type of data cluster is extracted. The relevant sample detection data included in the second type of data cluster has the same high-frequency sample oil and gas exploration scheme as the target sample detection data. The unrelated sample detection data included in the second type of data cluster belongs to at least one other original sample detection data besides the relevant sample detection dataset in the fourth type of data cluster and the relevant sample detection dataset in the second type of data cluster. The high-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on the sample oil and gas exploration scheme of each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is greater than the reference first number.

5. The oil and gas exploration method based on a smart detector as described in claim 1, characterized in that, The first number of comparative sample detection data clusters include a third type of data cluster and a fourth type of data cluster; The step of extracting the target sample detection data cluster includes: In a pre-set oil and gas exploration database, the target sample detection data and the initial exploration area features of the target sample detection data are extracted. The pre-set oil and gas exploration database includes at least two original sample detection data and the initial exploration area features of each original sample detection data. Based on the initial exploration area characteristics, the fourth type of data cluster is determined in the preset oil and gas exploration database. The data similarity between the relevant sample detection data included in the fourth type of data cluster and the initial exploration area characteristics is greater than or equal to the pre-configured reference data similarity, and they have the same low-frequency sample oil and gas exploration scheme or at least two identical sample oil and gas exploration schemes. The unrelated sample detection data included in the fourth type of data cluster belongs to at least one other original sample detection data outside the relevant sample detection dataset in the fourth type of data cluster. The low-frequency sample oil and gas exploration scheme belongs to the sample oil and gas exploration scheme calculated based on each original sample detection data in the preset oil and gas exploration database, whose occurrence frequency is less than the second reference number. In the preset oil and gas exploration database, the third type of data cluster is extracted. The relevant sample detection data included in the third type of data cluster has the same oil and gas exploration scheme distribution data as the target sample detection data. The unrelated sample detection data included in the third type of data cluster belongs to at least one other original sample detection data other than the relevant sample detection dataset in the fourth type of data cluster and the relevant sample detection dataset in the third type of data cluster.

6. The oil and gas exploration method based on a smart detector as described in claim 1, characterized in that, The step of calculating the corresponding scheme adaptation error parameters based on the fit distribution parameters of the detector data for each of the sample exploration areas and at least one sample oil and gas exploration scheme corresponding to the detector data for each of the sample exploration areas includes: For each undetermined oil and gas exploration scheme in the second number of sample oil and gas exploration schemes, the corresponding exploration scheme adaptation parameters are obtained based on the adaptation distribution parameters of the detector data of each sample exploration area. For each undetermined oil and gas exploration scheme in the second number of sample oil and gas exploration schemes, based on at least one sample oil and gas exploration scheme corresponding to the detector data of each sample exploration area, the corresponding actual adaptation parameters of the undetermined oil and gas exploration scheme are obtained. For each of the second number of sample oil and gas exploration schemes, the local scheme adaptation error parameter corresponding to the undetermined oil and gas exploration scheme is calculated based on the exploration scheme adaptation parameters and actual adaptation parameters of the detector data of each sample exploration area relative to the undetermined oil and gas exploration scheme. Based on the local scheme adaptation error parameters corresponding to each of the proposed oil and gas exploration schemes, the corresponding scheme adaptation error parameters are calculated.

7. The oil and gas exploration method based on a smart detector as described in claim 1, characterized in that, The step of calculating the corresponding detection contrast error parameter based on the sample exploration area characteristics of each sample exploration area detection data includes: Based on the first number of comparative sample detection data clusters and the target sample detection data, a first number of first training detection events and a first number of second training detection events are formed. Each first training detection event includes a related sample detection dataset of the comparative sample detection data cluster, and each second training detection event includes a related sample detection dataset and / or an unrelated sample detection dataset of the comparative sample detection data cluster. Based on the first number of first training detection events and the first number of second training detection events, a first number of local detection comparison error parameters are calculated. Based on the first number of local detection contrast error parameters, the corresponding detection contrast error parameters are calculated.

8. The oil and gas exploration method based on a smart geophone as described in any one of claims 1-7, characterized in that, The step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network to obtain the oil and gas feature mining network further includes: Identify at least two target sample detection data and a first number of comparison sample detection data clusters corresponding to each target sample detection data; Based on the detection data of the at least two target samples and the first number of comparison sample detection data clusters corresponding to each of the target sample detection data, a third training detection event and a fourth training detection event are formed. The data similarity level corresponding to the relevant sample detection data combination in the third training detection event is greater than the data similarity level corresponding to the unrelated sample detection data combination in the fourth training detection event. Based on the third training detection event and the fourth training detection event, the training comparison error parameter is calculated. The step of optimizing and adjusting the network parameters of the candidate oil and gas feature mining network based on the scheme adaptation error parameters and the detection comparison error parameters to form a corresponding oil and gas feature mining network includes: Based on the scheme adaptation error parameters, the detection contrast error parameters, and the training contrast error parameters, the network parameters of the candidate oil and gas feature mining network are optimized and adjusted to form a corresponding oil and gas feature mining network.

9. An oil and gas exploration system based on an intelligent geophone, characterized in that, It includes a processor and a memory, the memory being used to store a computer program, and the processor being used to execute the computer program to implement the oil and gas exploration method based on a smart detector as described in any one of claims 1-8.

Images