A method, apparatus, equipment, and storage medium for locating residual oil.

Abstract

This application discloses a method, apparatus, equipment, and storage medium for locating remaining oil. The method includes: acquiring segmented displacement state data of well groups and boundary transition trajectory data of the development stage; obtaining a continuous potential displacement state sequence by boundary interception, trajectory insertion, and time concatenation; obtaining a continuous displacement state sequence over all time periods by using a three-layer fully connected decoding method mirroring the encoding layer of the preceding depth Koopman autoencoder network; extracting the state at six consecutive sampling times at the boundary to form a local continuous change sequence; obtaining displacement obstruction judgment parameters through two-layer fully connected compression mapping; generating two types of markers through threshold comparison, cross-screening contradictory well groups, associating spatial locations, and merging to obtain the location result. This application can accurately distinguish two types of easily confused areas, and the location result can directly support the selection of sites and potential tapping decisions for later-stage oilfield development measures.

Description

Technical Field

[0001] This application relates to the field of oil and gas field development engineering technology, and in particular to a method, apparatus, equipment and storage medium for locating residual oil. Background Technology

[0002] As waterflooding development enters its mid-to-late stages, target well groups generally exhibit a coexistence of high water cut and declining oil production. Inter-well connectivity continues to evolve under long-term injection and production, and the distribution of remaining oil exhibits strong heterogeneity and time-varying characteristics. On-site production management typically organizes adjustments to the injection and production regime on a development phase-by-phase basis. Switching between development phases leads to a fundamental reorganization of the injection-production relationship and response propagation paths, causing the same well group to exhibit different dynamic patterns at different times.

[0003] The core of remaining oil utilization decisions is accurately identifying areas with potential for further oil displacement that are difficult to displace. Existing technologies for remaining oil identification mainly include numerical simulation, dynamic analysis, and data-driven methods. Numerical simulation requires the construction of detailed geological models, resulting in high computational costs and long cycles, and historical fitting exhibits multiple solutions. Dynamic analysis relies on engineering experience and struggles to quantitatively characterize complex displacement relationships between wells. Data-driven methods typically directly fit full-time time-series signals, treating development stage boundaries as ordinary time points, and cross-boundary state propagation often employs continuous extrapolation or interpolation smoothing, leading to excessive compression of the state propagation mechanism changes caused by injection-production relationship reorganization.

[0004] The external dynamics of areas where residual oil is difficult to displace are highly similar to those of high-wave and low-response areas after the transition during the development phase, both exhibiting characteristics of declining oil production and high water cut. Current technologies cannot distinguish between these two types of areas with completely different mechanisms, often misclassifying high-wave and low-response areas as areas rich in residual oil, resulting in a large amount of ineffective measures and severely impacting the potential tapping benefits in the later stages of oilfield development. Therefore, accurately distinguishing between areas where residual oil is difficult to displace and high-wave and low-response areas, and achieving reliable positioning of areas where residual oil is difficult to displace, is a critical problem that urgently needs to be solved. Summary of the Invention

[0005] In view of this, the residual oil positioning method, apparatus, device, and storage medium provided in this application can accurately distinguish between areas where residual oil is difficult to displace and areas with high amplitude and low response, thus achieving reliable positioning of areas where residual oil is difficult to displace. The residual oil positioning method, apparatus, device, and storage medium provided in this application are implemented as follows: This application provides a method for locating residual oil, including: Acquire well group segment displacement status data and development stage boundary transition trajectory data; The segmented displacement state data of the well group is processed by truncating the boundary before and after the boundary to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; the boundary transition trajectory data of the development stage is inserted between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; the continuous state segment and the remaining state are connected in chronological order to obtain a potential displacement state sequence of the continuous region. A three-layer fully connected decoding structure mirroring the encoding layer of the preceding deep Koopman autoencoder network is used to perform three fully connected mapping processes on each latent state in the potential displacement state sequence of the continuous region, to obtain the four-dimensional displacement state corresponding to each sampling time. The four-dimensional displacement states of all sampling times are integrated in chronological order to obtain the continuous displacement state sequence of the entire time period. Locate the position index of each development stage boundary in the full-time continuous displacement state sequence, extract the four-dimensional displacement state corresponding to 6 consecutive sampling times for each boundary position index, and obtain the local continuous change sequence; The local continuous change sequence is expanded into a local change vector, and the local change vector is compressed and mapped sequentially through a two-layer fully connected network to obtain the displacement resistance determination parameters. The displacement obstruction determination parameters are compared with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high wave and low response markers; the remaining oil difficult to continue displacement markers and the high wave and low response markers are cross-screened on a well group basis, and well groups that contain both the remaining oil difficult to continue displacement markers and the high wave and low response markers are removed; the screened well groups are associated with spatial locations, and adjacent well groups are merged to obtain the location results of the remaining oil difficult to continue displacement area.

[0006] In some embodiments, the well group segmented displacement state data is processed by truncating the boundary before and after the boundary to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; the development stage boundary transition trajectory data is inserted between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; the continuous state segment and the remaining state are connected in chronological order to obtain a continuous regional potential displacement state sequence, including: The displacement state data of the well group in segments are processed by truncation before and after the boundary to obtain the remaining state of the first segment and the remaining state of the second segment. The development phase boundary transition trajectory data, the corresponding remaining states of the previous segment and the remaining states of the subsequent segment are subjected to position matching and insertion processing to obtain continuous state segments; The continuous state segments and the remaining states are connected in a time sequence to obtain a sequence of potential displacement states in a continuous region.

[0007] In some embodiments, the three-layer fully connected decoding structure, mirroring the encoding layer of the preceding deep Koopman autoencoder network, is used to sequentially perform three fully connected mapping processes on each latent state in the continuous region latent displacement state sequence to obtain the four-dimensional displacement state corresponding to each sampling time; the four-dimensional displacement states of all sampling times are integrated in chronological order to obtain a continuous displacement state sequence for the entire time period, including: A first fully connected mapping process is performed on each potential state in the potential displacement state sequence of the continuous region to obtain the first decoding intermediate feature; The first decoded intermediate feature is subjected to a second fully connected mapping process to obtain the second decoded intermediate feature; The second decoded intermediate features are subjected to a third fully connected mapping process to obtain the displacement state corresponding to each sampling time. The displacement states corresponding to each sampling time are integrated in time sequence to obtain a continuous displacement state sequence for the entire time period.

[0008] In some embodiments, locating the position index of each development stage boundary in the full-time continuous displacement state sequence, and extracting the four-dimensional displacement state corresponding to six consecutive sampling times for each boundary position index to obtain a locally continuous change sequence includes: The boundary position index is obtained by performing localization processing on the continuous displacement state sequence throughout the entire time period; The continuous displacement states corresponding to the boundary position indices in the full-time continuous displacement state sequence are extracted to obtain a local continuous change sequence.

[0009] In some embodiments, the displacement obstruction determination parameter is compared with five preset threshold parameters to obtain residual oil displacement difficulty markers and high wave and low response markers; the residual oil displacement difficulty markers and the high wave and low response markers are cross-screened on a well group basis to remove well groups that simultaneously contain both residual oil displacement difficulty markers and high wave and low response markers; after associating the screened well groups with spatial locations, adjacent well groups are merged to obtain the location result of the residual oil displacement difficulty area, including: Five sets of threshold parameters are preset, namely, state change amplitude threshold, continuous decay length threshold, response hysteresis accumulation threshold, high wave and state change amplitude threshold, and high wave and continuous decay length threshold. If the state change amplitude before and after the boundary is less than the state change amplitude threshold, the continuous decay length after the boundary is greater than the continuous decay length threshold, and the response hysteresis accumulation is greater than the response hysteresis accumulation threshold, then a mark indicating that the remaining oil is difficult to continue to be displaced is generated. If the state change amplitude before and after the boundary is greater than the high wave and state change amplitude threshold, and the continuous decay length after the boundary is less than the high wave and continuous decay length threshold, then a mark indicating high wave and low response is generated. All markers are collected by target well group to obtain single well marker set; well groups that contain both markers indicating that the remaining oil is difficult to continue to be displaced and markers with high wave intensity and low response are removed to obtain regional screening results; The well groups in the regional screening results are mapped to their corresponding spatial locations and control ranges. Spatially adjacent well groups belonging to the same stratum are merged to obtain the location results of the remaining oil difficult to continue to be displaced, including boundary coordinates, control area and stratum.

[0010] In some embodiments, the step of expanding the local continuous change sequence into a local change vector, and then sequentially compressing and mapping the local change vector through a two-layer fully connected network to obtain displacement resistance determination parameters includes: The local continuous change sequence containing 6 four-dimensional states is unfolded in time order into a local change vector of length 24; The local change vector is compressed using a first fully connected layer to obtain a 6-dimensional feature vector. The 6-dimensional feature vector corresponds to the mean difference between the states before and after the boundary, the slope of the state change, the maximum fluctuation amplitude, the decay trend coefficient, the mean of the response lag, and the variance of the response lag, respectively. The 6-dimensional feature vector is processed by a second fully connected mapping to obtain the displacement obstruction determination parameters, which include the magnitude of state change before and after the development stage boundary, the length of continuous decay after the development stage boundary, and the cumulative amount of response hysteresis.

[0011] This application provides a residual oil positioning device, comprising: The acquisition module is used to acquire well group segment displacement status data and development stage boundary transition trajectory data; The processing module is used to perform boundary-cutting processing on the segmented displacement state data of the well group to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; to insert the boundary transition trajectory data of the development stage between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; and to connect the continuous state segment and the remaining state in chronological order to obtain a potential displacement state sequence of the continuous region. The processing module is further configured to perform three fully connected mapping processes on each latent state in the potential displacement state sequence of the continuous region using a three-layer fully connected decoding structure mirrored with the encoding layer of the preceding deep Koopman autoencoder network, to obtain the four-dimensional displacement state corresponding to each sampling time; and to integrate the four-dimensional displacement states of all sampling times in chronological order to obtain the continuous displacement state sequence of the entire time period. The processing module is also used to locate the position index of each development stage boundary in the full-time continuous displacement state sequence, extract the four-dimensional displacement state of each boundary position index for 6 consecutive sampling times, and obtain a local continuous change sequence. The processing module is further configured to expand the local continuous change sequence into a local change vector, and then compress and map the local change vector sequentially through a two-layer fully connected network to obtain the displacement resistance determination parameters. The positioning module is used to compare the displacement obstruction judgment parameters with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high wave and low response markers; to perform cross-screening processing on the remaining oil difficult to continue displacement markers and the high wave and low response markers on a well group basis, and to remove well groups that contain both the remaining oil difficult to continue displacement markers and the high wave and low response markers; after associating the screened well groups with spatial locations, adjacent well groups are merged to obtain the positioning results of the remaining oil difficult to continue displacement area.

[0012] The computer device provided in this application includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the method described in this application.

[0013] The computer-readable storage medium provided in this application embodiment stores a computer program thereon, which, when executed by a processor, implements the method described in this application embodiment.

[0014] The present application provides a method, apparatus, device, and storage medium for locating remaining oil. This method acquires segmented displacement state data of well groups and boundary transition trajectory data of the development stage. Through boundary interception, trajectory insertion, and time concatenation, a continuous potential displacement state sequence is obtained. A three-layer fully connected decoding method, mirroring the encoding layer of the preceding depth Koopman autoencoder network, is used to obtain a continuous displacement state sequence over all time periods. The state at six consecutive sampling times at the boundary is extracted to form a local continuous change sequence. Displacement obstruction judgment parameters are obtained through two-layer fully connected compression mapping. Two types of markers are generated through threshold comparison, and after cross-screening out contradictory well groups, their spatial locations are associated and merged to obtain the location result. This method can accurately distinguish between areas where remaining oil is difficult to displace and areas with high wave intensity and low response, achieving reliable location of areas where remaining oil is difficult to displace, thus solving the technical problems mentioned in the background art. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic diagram illustrating the implementation process of a method for locating residual oil provided in an embodiment of this application; Figure 2 A schematic diagram illustrating the implementation process of a potential displacement state sequence in a continuous region, as provided in an embodiment of this application; Figure 3 This is a schematic diagram of a residual oil positioning device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] The following description of some technologies involved in the embodiments of this application is provided to aid understanding and should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, some descriptions of well-known functions and structures are omitted in the following description.

[0019] Figure 1 This is a schematic flowchart illustrating the implementation of a method for locating residual oil according to an embodiment of this application, including steps 101 to 106. Wherein, Figure 1 This is merely one execution order shown in the embodiments of this application and does not represent the only execution order for a method of locating remaining oil. Where the final result can be achieved, Figure 1 The steps shown can be performed in parallel or in reverse order.

[0020] Step 101: Obtain well group segment displacement status data and development stage boundary transition trajectory data.

[0021] In this embodiment, the well group segmented displacement state data includes the regional potential displacement state sequence of the target well group and adjacent wells within each development stage segment. Each regional potential displacement state is an eight-dimensional vector, generated by the encoding layer of the pre-improved depth Koopman autoencoder network, representing the implicit state of the subsurface displacement system at the corresponding sampling time. The development stage boundary transition trajectory data is a continuous state sequence generated by the conditional diffusion bridge. Each boundary corresponds to 6 consecutive eight-dimensional states, namely the pre-boundary baseline state, 4 intermediate transition states, and the post-boundary baseline state, completely covering the state change process before and after the development stage switch. All of the above data are standardized outputs of the pre-module, requiring no additional downhole monitoring or geological modeling, and can be generated solely based on surface-collected injection volume, oil production, water cut, and inter-well response data.

[0022] Step 102: Perform boundary truncation processing on the segmented displacement state data of the well group to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; insert the boundary transition trajectory data of the development stage between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; connect the continuous state segment and the remaining state in chronological order to obtain a potential displacement state sequence of the continuous region.

[0023] In this embodiment, the potential displacement states of each development stage segment are first precisely extracted: all states before the pre-boundary reference state in the first development stage segment are extracted to obtain the remaining states of the first segment; all states after the post-boundary reference state in each intermediate development stage segment and before the next pre-boundary reference state are extracted to obtain the remaining states of the intermediate segments; and all states after the post-boundary reference state in the last development stage segment are extracted to obtain the remaining states of the final segment. This extraction method avoids the repetition of boundary endpoint states, ensuring that the remaining states of each segment contain only continuous evolution information within a single development stage. The transition trajectory corresponding to the boundary of each development stage is inserted between the end of the remaining state of the previous segment and the beginning of the remaining state of the next segment, so that the last state of the previous segment and the first state of the transition trajectory (pre-boundary reference state) are continuous in time index, and the last state of the transition trajectory (post-boundary reference state) and the first state of the next segment are continuous in time index, forming a continuous state segment corresponding to that boundary. Finally, in the order of the development stages, the remaining states of the first segment, all continuous state segments, and the remaining states of the last segment are connected in sequence to obtain a complete continuous region potential displacement state sequence covering the development from the start to the current time. The total length of the sequence is equal to the sum of the lengths of the remaining states of each segment and the sum of the lengths of all transition trajectories.

[0024] Step 103: A three-layer fully connected decoding structure mirroring the encoding layer of the preceding deep Koopman autoencoder network is used to perform three fully connected mapping processes on each latent state in the potential displacement state sequence of the continuous region in turn to obtain the four-dimensional displacement state corresponding to each sampling time. The four-dimensional displacement states of all sampling times are integrated in chronological order to obtain the continuous displacement state sequence of the entire time period.

[0025] In this embodiment, a three-layer fully connected decoding structure, completely mirroring the encoding layer of a deep Koopman autoencoder network, is adopted to ensure the reversibility and feature consistency of the encoding and decoding processes. The parameters of the decoding layer are completely shared with the parameters of the decoding layer of the preceding network, eliminating the need for retraining. For each eight-dimensional latent state in the continuous region's potential displacement state sequence, three fully connected mappings are performed sequentially: the first fully connected mapping expands the eight-dimensional features to thirty-two dimensions, the second fully connected mapping compresses the thirty-two-dimensional features to sixteen dimensions, and the third fully connected mapping restores the sixteen-dimensional features to a four-dimensional displacement state. The component order of the four-dimensional displacement state is completely consistent with the displacement feature sequence generated by the preceding module, namely, injection-production difference, water cut change rate, oil production decline, and response hysteresis, ensuring that the physical meaning of the decoding result is clear. The four-dimensional displacement states at all sampling times are arranged in time index order to obtain a continuous displacement state sequence for the entire time period. This sequence simultaneously contains stable evolution information within each development stage and transitional change information at the boundaries of development stages.

[0026] Step 104: Locate the position index of each development stage boundary in the continuous displacement state sequence throughout the entire time period, extract the four-dimensional displacement state corresponding to 6 consecutive sampling times for each boundary position index, and obtain the local continuous change sequence.

[0027] In this embodiment, a time index mapping table is first established between the development stage boundaries and the continuous displacement state sequence throughout the entire time period. Since the transition trajectory corresponding to each development stage boundary contains 6 sampling times, each boundary occupies 6 consecutive index positions in the potential displacement state sequence of the continuous region during continuous reconstruction. Based on this index mapping, the starting index of each development stage boundary in the continuous displacement state sequence throughout the entire time period is located, and this index corresponds to the decoding result of the baseline state before the boundary. Starting from this starting index, the four-dimensional displacement state corresponding to the next 6 sampling times is continuously extracted to form the local continuous change sequence corresponding to the boundary. This sequence corresponds exactly to the complete result after the transition trajectory is decoded, fully preserving the continuous change characteristics of the displacement state during the development stage switching process, and avoiding the loss of process information caused by only extracting the state of the endpoints before and after the boundary.

[0028] Step 105: Expand the local continuous change sequence into a local change vector, and then compress and map the local change vector sequentially through a two-layer fully connected network to obtain the displacement obstruction judgment parameters.

[0029] In this embodiment, the local continuous change sequence containing six four-dimensional states is first expanded into a one-dimensional vector in chronological order. The first four components correspond to the baseline state before the boundary, the next four components correspond to the first intermediate transition state, and so on, ultimately forming a local change vector of length 24. This vector is then input into a first fully connected compression layer containing six neurons. Through linear transformation and Rectified Linear Unit (ReLU) activation, the 24-dimensional high-dimensional features are compressed into six-dimensional statistics. The six-dimensional statistics correspond to: the difference in mean between states before and after the boundary, the slope of the state change, the maximum fluctuation amplitude, the decay trend coefficient, the mean of the response lag, and the variance of the response lag, comprehensively characterizing the displacement dynamic change features near the boundary of the development stage from different dimensions. The six-dimensional statistics are then input into a second fully connected mapping layer containing three neurons. Through linear transformation, the six-dimensional statistics are mapped into three-dimensional displacement resistance determination parameters. The three components are, in order, the amplitude of the state change before and after the boundary of the development stage, the length of the continuous decay after the boundary of the development stage, and the cumulative amount of the response lag.

[0030] Step 106: Compare the displacement obstruction judgment parameters with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high wave and low response markers; perform cross-screening on the remaining oil difficult to continue displacement markers and high wave and low response markers on a well group basis, and remove well groups that contain both the remaining oil difficult to continue displacement markers and high wave and low response markers; after associating the screened well groups with spatial locations, merge adjacent well groups to obtain the location results of the remaining oil difficult to continue displacement area.

[0031] In this embodiment, five sets of threshold parameters are first preset. All thresholds are obtained by statistical analysis of historical development data of the target oilfield block: the first threshold is the state change amplitude threshold, with a value range of 0.1 to 0.3; the second threshold is the continuous attenuation length threshold, with a value range of 3 to 6 sampling times; the third threshold is the response hysteresis accumulation threshold, with a value range of 2 to 5 sampling times; the fourth threshold is the high wave and state change amplitude threshold, with a value range of 0.5 to 0.8; and the fifth threshold is the high wave and continuous attenuation length threshold, with a value range of 1 to 2 sampling times. The 3D displacement obstruction determination parameters are compared one by one with the above thresholds: if the change in state before and after the boundary is less than the first threshold, the continuous attenuation length after the boundary is greater than the second threshold, and the cumulative response lag is greater than the third threshold, then the area corresponding to the boundary is determined to be obstructed, and a marker indicating that the remaining oil is difficult to continue to be displaced is generated; if the change in state before and after the boundary is greater than the fourth threshold and the continuous attenuation length after the boundary is less than the fifth threshold, then the area corresponding to the boundary is determined to be sufficiently affected, and a high-impact, low-response marker is generated; if neither of these conditions is met, then the displacement state of the area corresponding to the boundary is determined to be normal, and no marker is generated. Then, all markers are collected by target well group, and a cross-screening operation is performed to exclude well groups with contradictory displacement states. Finally, the screened well groups are associated with spatial locations, and adjacent well groups are merged to form continuous areas where the remaining oil is difficult to continue to be displaced.

[0032] This application embodiment continuously reconstructs the segmented displacement states and boundary transition trajectories to obtain a continuous sequence of potential displacement states covering the entire development cycle. It retains stable evolution information within each development stage and transitional change information at the development stage boundaries, providing complete process data support for determining displacement obstruction. Relying solely on standardized data output from the pre-module, no additional downhole monitoring, geological modeling, or manual interpretation is required. All input data can be obtained through conventional surface production monitoring, significantly reducing the cost of technology application and adapting to the production data conditions in the later stages of mature oilfield development. The final output of the remaining oil displacement area location results includes clear spatial coordinates, control range, and corresponding development stage information, which can be directly used for later-stage oilfield development measures, well group adjustments, and potential tapping scheme formulation, achieving a closed-loop delivery from data to decision.

[0033] In the above Figure 1 Based on the above, this application embodiment also provides a schematic diagram of the implementation process for obtaining a sequence of potential displacement states in a continuous region. For example... Figure 2 As shown, steps 201 to 203 are included: Step 201: Perform boundary interception processing on the segmented displacement state data of the well group to obtain the remaining state of the previous segment and the remaining state of the next segment.

[0034] In this embodiment, the potential displacement state sequence corresponding to the first development stage segment is located, the index position i corresponding to the baseline state before the boundary within the segment is found, and all potential displacement states from index 0 to i-1 are extracted to form the remaining state of the first segment. This part of the state completely represents the continuous displacement evolution process of the first development stage from the start to the transition before the development stage, and does not include any boundary endpoint states.

[0035] The potential displacement states in the region from the baseline state after the boundary of an intermediate development stage segment to the baseline state before the next boundary are truncated to obtain the remaining states of the intermediate segment. Specifically, for the k-th intermediate development stage segment (2≤k≤K, where K is the total number of development stages), the starting index j corresponding to the baseline state after the boundary of this segment and the ending index m corresponding to the baseline state before the boundary of the next development stage are located. All potential displacement states from index j+1 to m-1 are truncated to form the remaining states of the k-th intermediate segment. This ensures that each remaining state of an intermediate segment contains only continuous displacement states within a single development stage, completely excluding endpoint states between the two boundaries and avoiding state duplication.

[0036] Step 202: Perform position matching and insertion processing on the boundary transition trajectory data of the development stage, the corresponding remaining states of the previous segment and the remaining states of the subsequent segment to obtain continuous state segments.

[0037] In this embodiment, for the k-th development stage boundary, its corresponding 6x8 state transition trajectory is inserted sequentially between the end of the k-th segmented remaining state and the beginning of the (k+1)-th segmented remaining state. This makes the last state of the k-th segmented remaining state temporally continuous with the first state of the transition trajectory (pre-boundary reference state), and the last state of the transition trajectory (post-boundary reference state) temporally continuous with the first state of the (k+1)-th segmented remaining state, forming a continuous state segment of length (length of the k-th segmented remaining state + 6 + length of the (k+1)-th segmented remaining state).

[0038] Step 203: Perform time-sequential connection processing on the continuous state segments and the remaining states to obtain the potential displacement state sequence of the continuous region.

[0039] In this embodiment, following the chronological order of development stages, the remaining states of the first segment are connected first, then the continuous state segments corresponding to the boundaries of the first to the (K-1)th development stages are connected sequentially, and finally all potential displacement states (i.e., the remaining states of the final segment) after the baseline state following the boundary of the Kth development stage segment are connected. After the connections are completed, a unique continuous time index is assigned to each state, forming a complete continuous sequence of potential displacement states covering the entire development process from the start of development to the current moment.

[0040] This application's embodiments seamlessly connect the originally discrete segmented displacement states with continuous boundary transition trajectories through boundary truncation and transition trajectory insertion, forming a temporally continuous regional potential displacement state sequence, avoiding state breaks or repetitions. The complete insertion of the development stage boundary transition trajectory into the corresponding position ensures that the continuous state sequence contains the actual change path of the development stage switching process, rather than simple endpoint interpolation or extrapolation, laying a core foundation for subsequently using process characteristics to distinguish between the two types of regions. By accurately trunculating the remaining states outside the boundary endpoints in each segment, it ensures that each state appears only once in the continuous sequence, eliminating data redundancy and guaranteeing a strict correspondence between the time indices of each segment state and the transition trajectory. The adoption of unified truncation and insertion rules completely standardizes the continuous reconstruction process for different well groups and different development stages, ensuring a consistent format for the output continuous regional potential displacement state sequence, facilitating subsequent batch processing and automated inference.

[0041] In some embodiments, a three-layer fully connected decoding structure mirroring the encoding layer of the preceding deep Koopman autoencoder network is used to perform three fully connected mapping processes on each latent state in the continuous region latent displacement state sequence to obtain the four-dimensional displacement state corresponding to each sampling time; the four-dimensional displacement states of all sampling times are integrated in chronological order to obtain the continuous displacement state sequence for the entire time period, including: performing a first fully connected mapping process on each latent state in the continuous region latent displacement state sequence to obtain the first decoding intermediate feature.

[0042] Specifically, the first fully connected layer of the decoding layer contains 32 neurons, with a weight matrix of dimension 8×32 and a bias vector of dimension 32. For each eight-dimensional region's potential displacement state h, a linear transformation result is obtained by calculating h×W1+b1, where W1 is the weight matrix of the first fully connected layer and b1 is the bias vector of the first fully connected layer. This result is then processed by the modified linear nonlinear activation function ReLU to obtain the 32-dimensional first decoding intermediate feature. The ReLU activation function can effectively avoid the gradient vanishing problem and improve the stability of the decoding process.

[0043] Furthermore, a second fully connected mapping process is performed on the first decoding intermediate feature to obtain the second decoding intermediate feature.

[0044] Specifically, the second fully connected layer of the decoding layer contains 16 neurons, with a weight matrix of dimension 32×16 and a bias vector of dimension 16. For the 32-dimensional first decoding intermediate feature f1, a linear transformation is obtained by calculating f1×W2+b2, where W2 is the weight matrix of the second fully connected layer and b2 is the bias vector of the second fully connected layer. This transformation is then processed by the ReLU activation function to obtain the 16-dimensional second decoding intermediate feature. This layer achieves progressive compression of the feature dimension, removing redundant information.

[0045] Furthermore, a third fully connected mapping process is performed on the intermediate features of the second decoding to obtain the displacement state corresponding to each sampling time.

[0046] Specifically, the third fully connected layer of the decoding layer contains four neurons, with a weight matrix of dimension 16×4 and a bias vector of dimension 4. For the 16-dimensional second decoding intermediate feature f2, a linear transformation is obtained by calculating f2×W3+b3, where W3 is the weight matrix of the first fully connected layer and b3 is the bias vector of the first fully connected layer. This layer does not use an activation function and directly outputs the four-dimensional displacement state. The four components are, in order, the injection-production difference, the water cut change rate, the oil production decline, and the response hysteresis, which are completely consistent with the component order of the displacement feature sequence in the preceding module.

[0047] Furthermore, the displacement states corresponding to each sampling time are integrated in time sequence to obtain a continuous displacement state sequence for the entire time period.

[0048] Specifically, the four-dimensional displacement states corresponding to all sampling times are arranged sequentially according to their time index in the continuous region potential displacement state sequence, forming a full-time continuous displacement state sequence with the number of rows equal to the total number of sampling times and the number of columns equal to 4. Each state in this sequence has a unique timestamp, fully representing the continuous change process of underground displacement dynamics from the start of development to the current moment.

[0049] This application employs a three-layer fully connected decoding structure that is a complete mirror image of the encoding layer of the pre-processed deep Koopman autoencoder network, with fully shared parameters. This ensures that the potential states can be accurately reconstructed into original displacement features with clear physical meaning, avoiding information distortion during the decoding process. The four-dimensional displacement state output by decoding is completely consistent with the component order of the displacement feature sequence generated by the pre-processing module, corresponding sequentially to injection-production difference, water cut change rate, oil production decline, and response hysteresis. Each component has a clear engineering physical meaning, facilitating subsequent analysis and verification. Each potential state in the potential displacement state sequence of a continuous region is decoded independently, ensuring the temporal continuity of the decoded continuous displacement state sequence and fully preserving the dynamic changes of the displacement process. The three-layer fully connected structure has low computational complexity and fast inference speed, and the ReLU activation function effectively avoids the gradient vanishing problem, ensuring the stability and reliability of the decoding process, making it suitable for batch processing of large-scale well groups.

[0050] In some embodiments, locating the position index of each development stage boundary in the full-time continuous displacement state sequence, extracting the four-dimensional displacement state corresponding to 6 consecutive sampling times for each boundary position index, and obtaining a local continuous change sequence includes: performing localization processing on the full-time continuous displacement state sequence to obtain the boundary position index.

[0051] Specifically, during continuous reconstruction, the starting index of the transition trajectory corresponding to the boundary of each development stage is recorded in the potential displacement state sequence of the continuous region. Since the time index remains unchanged during the decoding process, this index is the starting index of the boundary in the continuous displacement state sequence of the entire time period, and is used as the boundary position index of the boundary of that development stage. A one-to-one correspondence between boundary numbers and boundary position indices is established to form a boundary position index table.

[0052] Furthermore, the continuous displacement states corresponding to the boundary position indices in the continuous displacement state sequence over the entire time period are extracted and processed to obtain a local continuous change sequence.

[0053] Specifically, for each development stage boundary, based on its boundary position index 's', six consecutive four-dimensional displacement states are extracted from the continuous displacement state sequence throughout the entire time period, including the current index and the next five indices. These six states correspond precisely to the complete sequence after decoding the transition trajectory of the development stage boundary, namely, the pre-boundary baseline state, the first intermediate transition state, the second intermediate transition state, the third intermediate transition state, the fourth intermediate transition state, and the post-boundary baseline state. Arranging these six states in chronological order forms a 6-row, 4-column local continuous change sequence corresponding to the boundary of that development stage.

[0054] This application's embodiments establish a time index mapping relationship between development phase boundaries and continuous displacement state sequences throughout the entire time period. This accurately locates the starting position corresponding to each development phase boundary, avoiding subsequent feature extraction deviations caused by boundary positioning errors. Extracting the continuous displacement states corresponding to the boundary positions fully encompasses all transitional change information before and after the development phase switch, rather than just extracting the endpoint states before and after the boundary. This solves the problem of indistinguishable two types of regions due to the loss of process information in existing technologies. The length of the extracted local continuous change sequence is completely consistent with the length of the development phase boundary transition trajectory, ensuring that the extracted features can fully characterize the entire boundary transition process, providing a comprehensive data foundation for subsequent feature compression and judgment. Extracting local change information near the boundary separately from the full-time sequence significantly reduces the amount of data required for subsequent processing, while enabling subsequent feature compression and judgment to focus on the most discriminative boundary transition region, improving processing efficiency and judgment accuracy.

[0055] In some embodiments, the displacement obstruction determination parameter is compared with five preset groups of threshold parameters to obtain a remaining oil difficult-to-displace mark and a high sweep and low response mark; the remaining oil difficult-to-displace mark and the high sweep and low response mark are cross-screened and processed in units of well groups, and the well groups that simultaneously contain the remaining oil difficult-to-displace mark and the high sweep and low response mark are excluded; after associating the screened well groups with spatial positions, adjacent well groups are merged to obtain the positioning result of the remaining oil difficult-to-displace area, including: five preset groups of threshold parameters, which are respectively the state change amplitude threshold, the continuous attenuation length threshold, the response lag cumulative amount threshold, the high sweep state change amplitude threshold, and the high sweep continuous attenuation length threshold; if the state change amplitude before and after the boundary in the displacement obstruction determination parameter is less than the state change amplitude threshold, the continuous attenuation length after the boundary is greater than the continuous attenuation length threshold, and the response lag cumulative amount is greater than the response lag cumulative amount threshold, a remaining oil difficult-to-displace mark is generated; if the state change amplitude before and after the boundary is greater than the high sweep state change amplitude threshold and the continuous attenuation length after the boundary is less than the high sweep continuous attenuation length threshold, a high sweep and low response mark is generated.

[0056] Specifically, five groups of threshold parameters are preset and obtained by statistically analyzing the historical development data of the target oilfield block in the past 10 years: the first threshold T1 = 0.2, the second threshold T2 = 4 sampling times, the third threshold T3 = 3 sampling times, the fourth threshold T4 = 0.6, and the fifth threshold T5 = 2 sampling times. For the 3D displacement obstruction determination parameter [b1, b2, b3], if b1 < T1 and b2 > T2 and b3 > T3, a remaining oil difficult-to-displace mark is generated at the boundary position in this development stage; if b1 > T4 and b2 < T5, a high sweep and low response mark is generated; if neither of the two conditions is met, no mark is generated.

[0057] Further, all marks are collected in units of the target well group to obtain a single-well mark set; the well groups that simultaneously contain the remaining oil difficult-to-displace mark and the high sweep and low response mark are excluded to obtain the regional screening result.

[0058] Specifically, in units of the target well group, all marks at the boundaries of all development stages are traversed, and all marks of the same well group are collected together to form a single-well mark set. The type of the single-well mark set is judged, and cross-screening operations are performed to exclude the well groups with contradictory displacement states and retain the well groups with clear potential tapping value.

[0059] Further, the well groups in the regional screening result are mapped to the corresponding spatial positions and control ranges, and the well groups that are adjacent in space and belong to the same formation are merged to obtain the positioning result of the remaining oil difficult-to-displace area including the boundary coordinates, the controlled area, and the formation to which it belongs.

[0060] Specifically, a mapping table is established between target well group numbers and their center coordinates, controlled area, and stratigraphic affiliation. The target well group numbers in the regional screening results are mapped to their corresponding spatial locations and controlled areas. Spatially adjacent target well groups belonging to the same stratigraphic layer are merged to form continuous areas of remaining oil that are difficult to displace; isolated target well groups are separately designated as independent areas of remaining oil that are difficult to displace. Finally, the boundary coordinates, included well group numbers, controlled area, stratigraphic affiliation, and corresponding development stage boundary index of each area of ​​remaining oil that is difficult to displace are output, forming a complete location result report.

[0061] This application's embodiments, through comparison of multiple sets of threshold parameters, simultaneously generate markers for remaining oil that is difficult to displace and markers for high wave intensity and low response. This allows for the mechanistic differentiation of two types of reservoir areas that appear similar but are fundamentally different, fundamentally avoiding the ineffective investment of misclassifying high wave intensity and low response areas as remaining oil-rich areas. By cross-screening the two types of markers, well groups with contradictory displacement states are excluded, further reducing the probability of misjudgment and ensuring the accuracy and reliability of the final area screening results. The area screening results are correlated with the spatial correspondence of the target well groups, transforming abstract markers into remaining oil areas with clear spatial locations and control ranges, enabling the positioning results to be directly applied to potential tapping operations in the oilfield. The preset multiple sets of threshold parameters can be flexibly adjusted according to the geological conditions and development history of different oilfield blocks, making this method adaptable to different types of reservoirs and possessing broad versatility.

[0062] In some embodiments, the local continuous change sequence is expanded into a local change vector, and the local change vector is compressed and mapped sequentially through a two-layer fully connected network to obtain the displacement obstruction determination parameters, including: expanding the local continuous change sequence containing 6 four-dimensional states into a local change vector of length 24 in chronological order.

[0063] Specifically, the six four-dimensional states in the local continuous change sequence are concatenated sequentially in chronological order. The four components of the first state are used as the first to fourth positions of the vector, the four components of the second state are used as the fifth to eighth positions, and so on, with the four components of the sixth state used as the 21st to 24th positions, ultimately forming a one-dimensional local change vector of length 24.

[0064] Furthermore, the local change vector is compressed using the first layer of fully connected processing to obtain a 6-dimensional feature vector. The 6-dimensional feature vector corresponds to the difference in mean state before and after the boundary, the slope of state change, the maximum fluctuation amplitude, the decay trend coefficient, the mean of response lag, and the variance of response lag, respectively.

[0065] Specifically, the first fully connected compressed layer contains 6 neurons, with a weight matrix dimension of 24×6 and a bias vector dimension of 6. For the 24-dimensional local change vector v, a linear transformation is obtained by calculating v×W4+b4, where W4 is the weight matrix of the first fully connected compressed layer and b4 is the bias vector. This transformation is then processed by the ReLU activation function to obtain a 6-dimensional statistic. The physical meanings of the 6-dimensional statistic are as follows: the first dimension is the difference in mean states before and after the boundary, representing the overall state difference on both sides of the boundary; the second dimension is the slope of the state change, representing the speed of the state change; the third dimension is the maximum fluctuation amplitude, representing the severity of the state change; the fourth dimension is the decay trend coefficient, representing the decay trend of the state after the boundary; the fifth dimension is the mean response lag, representing the average lag time of the inter-well response; and the sixth dimension is the variance of the response lag, representing the dispersion of the inter-well response lag.

[0066] Furthermore, a second fully connected mapping process is performed on the 6-dimensional feature vector to obtain the displacement obstruction determination parameters, which include the magnitude of state change before and after the development stage boundary, the length of continuous decay after the development stage boundary, and the cumulative amount of response hysteresis.

[0067] Specifically, the second fully connected mapping layer contains 3 neurons, with a weight matrix of dimension 6×3 and a bias vector of dimension 3. For the 6-dimensional statistic s, a linear transformation is obtained by calculating s×W5+b5, where W5 is the weight matrix of the second fully connected mapping layer and b5 is the bias vector. This layer does not use an activation function and directly outputs 3-dimensional displacement resistance determination parameters. The three components are, in order: the magnitude of state change before and after the development phase boundary, the duration of sustained decay after the development phase boundary, and the cumulative response hysteresis.

[0068] This embodiment employs a two-layer fully connected processing method to progressively compress a high-dimensional local change vector containing the complete boundary transition process into low-dimensional displacement resistance determination parameters. This removes redundant noise information and extracts the most discriminative core features. The output three-dimensional displacement resistance determination parameters correspond to the magnitude of state changes before and after the boundary in the development stage, the length of sustained decay after the boundary, and the cumulative response hysteresis. Each parameter has a clear engineering and physical meaning, facilitating understanding and verification by engineers. Quantifying the complex boundary change process into three simple determination parameters significantly simplifies the subsequent threshold comparison process. Furthermore, since the parameters are core features extracted from the complete transition process, the accuracy of the determination results is guaranteed. The two-layer fully connected structure has a small number of parameters, low training difficulty, and requires only a small number of labeled samples for training. The trained model also exhibits fast inference speed, meeting the real-time processing needs of oilfields.

[0069] In some embodiments, the remaining oil difficult to displace markers and high wave and low response markers are cross-screened to obtain regional screening results, including: collecting all markers by target well group to obtain single well marker set.

[0070] Specifically, all markers generated at the boundary locations of all development stages are traversed. Each marker contains three fields: well group number, boundary number, and marker type. All markers are grouped according to the well group number, and all remaining oil difficult to displace and high wave and low response markers generated at the boundaries of different development stages of the same well group are aggregated together to form the single well marker set corresponding to that well group.

[0071] Furthermore, target well groups that simultaneously contain markers indicating that remaining oil is difficult to displace and markers indicating high wave and low response are eliminated.

[0072] Specifically, if a target well group has at least one marker indicating that the remaining oil is difficult to displace and at least one marker indicating high wave intensity and low response, then the displacement status of the well group is considered to be contradictory. This may be due to the extremely strong heterogeneity within the well group or frequent adjustments to the development system, making it impossible to clearly determine that it is an area where the remaining oil is difficult to displace. To avoid misjudgment, the target well group is removed from the candidate area.

[0073] Furthermore, target well groups containing only remaining oil that is difficult to displace are retained to obtain regional screening results.

[0074] Specifically, if the set of markers for a target well group contains only markers indicating remaining oil that is difficult to displace, and does not contain any high-wave or low-response markers, then the area corresponding to that well group is determined to be an area with clear potential for further oil displacement, and the target well group is retained in the regional screening results. If the set of markers for a well group contains only high-wave and low-response markers, then the well group is determined to have no potential for further oil displacement and is not included in the regional screening results.

[0075] This application's implementation fundamentally solves the problem of misjudgment in two types of easily confused areas through a well group-level marker aggregation and cross-screening mechanism. First, it aggregates markers from all development stage boundaries at the target well group level, forming a single-well marker set. This avoids the one-sidedness of judging a single boundary and comprehensively reflects the overall displacement status of the well group. Second, cross-screening eliminates contradictory well groups containing both types of markers, ruling out uncertainties caused by strong heterogeneity within the well group or frequent adjustments to the development system, significantly reducing the risk of misjudgment. Finally, it retains only well groups containing markers indicating that remaining oil is difficult to continue displacement, ensuring that the screening results are all areas with clear potential for development, while automatically excluding high-volume and low-response well groups with no potential for development. This scheme is logically clear, highly interpretable, effectively avoids ineffective measures, and significantly improves the potential tapping benefits in the later stages of oilfield development.

[0076] While this application provides the method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in this embodiment is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the methods shown in this embodiment or the accompanying drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0077] like Figure 3 As shown in the illustration, this application also provides a residual oil positioning device 300. The device includes: The acquisition module 301 is used to acquire well group segment displacement state data and development stage boundary transition trajectory data.

[0078] Processing module 302 is used to perform boundary interception processing on the segmented displacement state data of the well group to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; to insert the boundary transition trajectory data of the development stage between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; and to connect the continuous state segment and the remaining state in chronological order to obtain a potential displacement state sequence of the continuous region.

[0079] The processing module 302 is also used to perform three fully connected mapping processes on each potential state in the potential displacement state sequence of the continuous region in a three-layer fully connected decoding structure mirrored with the coding layer of the preceding deep Koopman autoencoder network, to obtain the four-dimensional displacement state corresponding to each sampling time; and to integrate the four-dimensional displacement states of all sampling times in chronological order to obtain the continuous displacement state sequence of the entire time period.

[0080] The processing module 302 is also used to locate the position index of each development stage boundary in the continuous displacement state sequence of the whole time period, extract the four-dimensional displacement state corresponding to 6 consecutive sampling times of each boundary position index, and obtain the local continuous change sequence.

[0081] The processing module 302 is also used to expand the local continuous change sequence into a local change vector, and to compress and map the local change vector sequentially through a two-layer fully connected network to obtain the displacement obstruction judgment parameters.

[0082] The positioning module 303 is used to compare the displacement obstruction judgment parameters with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high wave and low response markers; cross-screening is performed on the remaining oil difficult to continue displacement markers and high wave and low response markers on a well group basis to remove well groups that contain both remaining oil difficult to continue displacement markers and high wave and low response markers; after associating the screened well groups with spatial locations, adjacent well groups are merged to obtain the positioning results of the remaining oil difficult to continue displacement area.

[0083] Some modules in the apparatus described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0084] The apparatus or module described in the above embodiments can be implemented by a computer chip or physical entity, or by a product with a certain function. For ease of description, the above apparatus is described by dividing it into various modules according to their functions. When implementing the embodiments of this application, the functions of each module can be implemented in one or more software and / or hardware. Of course, a module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0085] The methods, apparatus, or modules described in this application can be implemented in a computer-readable program code manner. The controller can be implemented in any suitable manner, such as a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of a memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code manner, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included within it for implementing various functions can also be considered as structures within the hardware component. Alternatively, the device used to implement various functions can be viewed as either a software module implementing the method or a structure within a hardware component.

[0086] This application also provides an apparatus, the apparatus comprising: a processor; a memory for storing processor-executable instructions; wherein, when the processor executes the executable instructions, it implements the method described in this application.

[0087] This application also provides a non-volatile computer-readable storage medium storing a computer program or instructions thereon, which, when executed, enables the method described in this application embodiment to be implemented.

[0088] Furthermore, in the various embodiments of the present invention, each functional module can be integrated into a processing module, or each module can exist independently, or two or more modules can be integrated into a single module.

[0089] The aforementioned storage media include, but are not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions.

[0090] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, or it can be embodied in the process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0091] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this application can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0092] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A method for locating residual oil, characterized in that, include: Acquire well group segment displacement status data and development stage boundary transition trajectory data; The segmented displacement state data of the well group is processed by truncating the boundary before and after the boundary to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; the boundary transition trajectory data of the development stage is inserted between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; the continuous state segment and the remaining state are connected in chronological order to obtain a potential displacement state sequence of the continuous region. A three-layer fully connected decoding structure mirroring the encoding layer of the preceding deep Koopman autoencoder network is used to perform three fully connected mapping processes on each latent state in the potential displacement state sequence of the continuous region, to obtain the four-dimensional displacement state corresponding to each sampling time. The four-dimensional displacement states of all sampling times are integrated in chronological order to obtain the continuous displacement state sequence of the entire time period. Locate the position index of each development stage boundary in the full-time continuous displacement state sequence, extract the four-dimensional displacement state corresponding to 6 consecutive sampling times for each boundary position index, and obtain the local continuous change sequence; The local continuous change sequence is expanded into a local change vector, and the local change vector is compressed and mapped sequentially through a two-layer fully connected network to obtain the displacement resistance determination parameters. The displacement obstruction determination parameters are compared with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high wave and low response markers; the remaining oil difficult to continue displacement markers and the high wave and low response markers are cross-screened on a well group basis to remove well groups that contain both the remaining oil difficult to continue displacement markers and the high wave and low response markers; the screened well groups are associated with spatial locations and adjacent well groups are merged to obtain the location results of the remaining oil difficult to continue displacement area; The step of comparing the displacement obstruction determination parameters with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high sweep and low response markers includes: Five sets of threshold parameters are preset, namely, state change amplitude threshold, continuous decay length threshold, response hysteresis accumulation threshold, high wave and state change amplitude threshold, and high wave and continuous decay length threshold. If the state change amplitude before and after the boundary is less than the state change amplitude threshold, the continuous decay length after the boundary is greater than the continuous decay length threshold, and the response hysteresis accumulation is greater than the response hysteresis accumulation threshold, then a mark indicating that the remaining oil is difficult to continue to be displaced is generated. If the state change amplitude before and after the boundary is greater than the high wave and state change amplitude threshold, and the continuous decay length after the boundary is less than the high wave and continuous decay length threshold, then a mark indicating high wave and low response is generated.

2. The method according to claim 1, characterized in that, The well group segmented displacement state data is processed by truncating the boundary before and after the boundary to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; the development stage boundary transition trajectory data is inserted between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; the continuous state segment and the remaining state are connected in chronological order to obtain a continuous region potential displacement state sequence, including: The displacement state data of the well group in segments are processed by truncation before and after the boundary to obtain the remaining state of the first segment and the remaining state of the second segment. The development phase boundary transition trajectory data, the corresponding remaining states of the previous segment and the remaining states of the subsequent segment are subjected to position matching and insertion processing to obtain continuous state segments; The continuous state segments and the remaining states are connected in a time sequence to obtain a sequence of potential displacement states in a continuous region.

3. The method according to claim 1, characterized in that, The three-layer fully connected decoding structure, which is mirrored by the encoding layer of the preceding deep Koopman autoencoder network, is used to perform three fully connected mapping processes on each potential state in the sequence of potential displacement states in the continuous region, so as to obtain the four-dimensional displacement state corresponding to each sampling time. By integrating the four-dimensional displacement states at all sampling times in chronological order, a continuous displacement state sequence for the entire time period is obtained, including: A first fully connected mapping process is performed on each potential state in the potential displacement state sequence of the continuous region to obtain the first decoding intermediate feature; The first decoded intermediate feature is subjected to a second fully connected mapping process to obtain the second decoded intermediate feature; The second decoded intermediate features are subjected to a third fully connected mapping process to obtain the displacement state corresponding to each sampling time. The displacement states corresponding to each sampling time are integrated in time sequence to obtain a continuous displacement state sequence for the entire time period.

4. The method according to claim 1, characterized in that, The process involves locating the position indices of the boundaries of each development stage in the continuous displacement state sequence over the entire time period, extracting the four-dimensional displacement state corresponding to six consecutive sampling times for each boundary position index, and obtaining a locally continuous change sequence, including: The boundary position index is obtained by performing localization processing on the continuous displacement state sequence throughout the entire time period; The continuous displacement states corresponding to the boundary position indices in the full-time continuous displacement state sequence are extracted to obtain a local continuous change sequence.

5. The method according to claim 1, characterized in that, The method involves cross-screening the remaining oil displacement markers and the high wave and low response markers on a well group basis, and eliminating well groups that simultaneously contain both the remaining oil displacement markers and the high wave and low response markers. After associating the selected well groups with spatial locations, adjacent well groups are merged to obtain the location results of areas where the remaining oil is difficult to displace, including: All markers are collected by target well group to obtain single well marker set; well groups that contain both markers indicating that the remaining oil is difficult to continue to be displaced and markers with high wave intensity and low response are removed to obtain regional screening results; The well groups in the regional screening results are mapped to their corresponding spatial locations and control ranges. Spatially adjacent well groups belonging to the same stratum are merged to obtain the location results of the remaining oil difficult to continue to be displaced, including boundary coordinates, control area and stratum.

6. The method according to claim 1, characterized in that, The step of expanding the local continuous change sequence into a local change vector, and then compressing and mapping the local change vector sequentially through a two-layer fully connected network to obtain the displacement resistance determination parameters includes: The local continuous change sequence containing 6 four-dimensional states is unfolded in time order into a local change vector of length 24; The local change vector is compressed using a first fully connected layer to obtain a 6-dimensional feature vector. The 6-dimensional feature vector corresponds to the mean difference between the states before and after the boundary, the slope of the state change, the maximum fluctuation amplitude, the decay trend coefficient, the mean of the response lag, and the variance of the response lag, respectively. The 6-dimensional feature vector is processed by a second fully connected mapping to obtain the displacement obstruction determination parameters, which include the magnitude of state change before and after the development stage boundary, the length of continuous decay after the development stage boundary, and the cumulative amount of response hysteresis.

7. A device for locating residual oil, characterized in that, include: The acquisition module is used to acquire well group segment displacement status data and development stage boundary transition trajectory data; The processing module is used to perform boundary-cutting processing on the segmented displacement state data of the well group to obtain the remaining state of the previous segment and the remaining state of the subsequent segment; to insert the boundary transition trajectory data of the development stage between the corresponding remaining state of the previous segment and the remaining state of the subsequent segment to obtain a continuous state segment; and to connect the continuous state segment and the remaining state in chronological order to obtain a potential displacement state sequence of the continuous region. The processing module is further configured to perform three fully connected mapping processes on each latent state in the potential displacement state sequence of the continuous region using a three-layer fully connected decoding structure mirrored with the encoding layer of the preceding deep Koopman autoencoder network, to obtain the four-dimensional displacement state corresponding to each sampling time; and to integrate the four-dimensional displacement states of all sampling times in chronological order to obtain the continuous displacement state sequence of the entire time period. The processing module is also used to locate the position index of each development stage boundary in the full-time continuous displacement state sequence, extract the four-dimensional displacement state of each boundary position index for 6 consecutive sampling times, and obtain a local continuous change sequence. The processing module is further configured to expand the local continuous change sequence into a local change vector, and then compress and map the local change vector sequentially through a two-layer fully connected network to obtain the displacement resistance determination parameters. The positioning module is used to compare the displacement obstruction judgment parameters with five preset threshold parameters to obtain the remaining oil difficult to continue displacement markers and high wave and low response markers; to perform cross-screening processing on the remaining oil difficult to continue displacement markers and the high wave and low response markers on a well group basis, and to remove well groups that contain both the remaining oil difficult to continue displacement markers and the high wave and low response markers; after associating the screened well groups with spatial locations, adjacent well groups are merged to obtain the positioning results of the remaining oil difficult to continue displacement area; The processing module is further configured to compare the displacement obstruction determination parameters with five preset threshold parameters to obtain a residual oil displacement difficulty marker and a high sweep and low response marker, wherein: Five sets of threshold parameters are preset, namely, state change amplitude threshold, continuous decay length threshold, response hysteresis accumulation threshold, high wave and state change amplitude threshold, and high wave and continuous decay length threshold. If the state change amplitude before and after the boundary is less than the state change amplitude threshold, the continuous decay length after the boundary is greater than the continuous decay length threshold, and the response hysteresis accumulation is greater than the response hysteresis accumulation threshold, then a mark indicating that the remaining oil is difficult to continue to be displaced is generated. If the state change amplitude before and after the boundary is greater than the high wave and state change amplitude threshold, and the continuous decay length after the boundary is less than the high wave and continuous decay length threshold, then a mark indicating high wave and low response is generated.

8. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images