Shale oil reservoir recovery ratio prediction method and device and readable storage medium

Abstract

The invention provides a shale oil reservoir recovery ratio prediction method and device and a readable storage medium, and relates to the technical field of shale oil reservoirs, and the method specifically comprises the steps: collecting multi-source data including geology, fluid properties and development process parameters in continuous T historical time periods, carrying out the deep analysis of parameters such as a porosity mean value, a permeability mean value, and the like, and carrying out the prediction of the recovery ratio of the shale oil reservoirs; the method comprises the following steps: constructing a sample feature matrix, training a recovery ratio change identification model, accurately judging the change condition of the recovery ratio, collecting multi-source data of an oil reservoir to be predicted to form a feature matrix input model, obtaining a recovery ratio change result and a corresponding average recovery ratio, and determining a reference sample through similarity fitting. According to the method, multi-source data are comprehensively collected, deep mining analysis is carried out, the defects that in the prior art, data collection is one-sided and analysis is shallow are overcome, the model can deal with complex oil reservoir conditions, the recovery ratio change result is combined with a reference sample, the one-sidedness of single-factor prediction is avoided, and the accuracy of shale oil reservoir recovery ratio prediction is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of shale reservoirs, and specifically provides a method and device for predicting the recovery rate of shale reservoirs and a readable storage medium. Background Art

[0002] With the continuous growth of global energy demand and the gradual scarcity of traditional oil resources, shale oil, as an important unconventional oil and gas resource, has attracted increasing attention in its development and utilization. Shale reservoirs have significantly different geological characteristics and production characteristics from conventional reservoirs. Geologically, the porosity and permeability of shale are extremely low, making it extremely difficult for oil to flow in the rock; its complex micro-porous structure, including the widespread existence of nano-scale pores, further increases the complexity of oil and gas migration. In terms of production, the production of shale oil usually requires special technical means, such as large-scale fracturing technology, to form an artificial fracture network and improve the oil flow channel.

[0003] In the prior art, a method for predicting the recovery rate of a fractured shale reservoir with the publication number of CN117634921A includes the following steps: collecting reservoir data information: measuring the oil storage volume of the fractured shale reservoir through geological exploration; constructing an underground geological model by collecting the underground structure, oil storage areas, etc. of the underground fractured shale, so as to obtain an overall view of the underground reservoir, dividing the collected samples by layer for layer tests, and calculating the recovery rates of different shale layers underground; for different oil storage areas in different layers, dividing them again through the constructed model, and calculating the different recovery rates of the oil storage areas existing in different areas of different layers, and the overall recovery rate of the reservoir can be obtained by combination; by adopting two calculation methods to predict the recovery rate, the method can perform combined prediction in the process of predicting the recovery rate of the fractured shale reservoir, with high prediction accuracy and simple and convenient prediction process.

[0004] However, there are still the following deficiencies. As can be seen from the above description, the prior art only relies on limited data obtained from geological exploration, such as oil storage volume, reservoir structure, and oil storage areas in rock layers, and does not fully explore other factors that may affect the recovery rate, such as rock physical properties, fluid characteristics, and production process parameters. Moreover, for the collected data, no in-depth analysis and excavation are carried out, and it is only simply used to construct a model and calculate the oil volume and recovery rate, resulting in insufficient understanding of the complex physical processes and interactions inside the reservoir. This makes the prediction model unable to accurately capture the non-linear relationship between the recovery rate and various factors, and when facing the dynamic changes of various factors in the actual production process, the prediction results often deviate greatly from the actual situation; at the same time, the area division and recovery rate calculation based on the 1:1 underground reservoir model rely too much on the accuracy of the model, and the actual reservoir situation is complex and changeable, and the model is difficult to fully and accurately reflect the real situation, resulting in limited accuracy.

[0005] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The object of the present invention is to provide a method, device and readable storage medium for predicting the recovery rate of shale oil reservoirs, so as to solve the problems raised in the above background art.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for predicting the recovery rate of shale oil reservoirs, the specific steps include:

[0009] S1. Collect multi-source data of a sample in consecutive T historical time periods. The sample includes shale oil reservoirs with unchanged recovery rate and those with changed recovery rate. Based on the shale oil reservoirs with unchanged recovery rate and those with changed recovery rate, calculate their average recovery rates in T historical time periods respectively. The multi-source data includes geological parameters, fluid property parameters and development process parameters;

[0010] S2. Extract characteristic parameters from the collected geological parameters, fluid property parameters and development process parameters respectively. Based on the extracted characteristic parameters, construct a characteristic matrix of each sample in consecutive T historical time periods;

[0011] S3. Construct a recovery rate change recognition model. Use the characteristic matrix of the sample in T historical time periods as the input, and use the recovery rate change result and the corresponding average recovery rate as labels to train the model. Train the recovery rate change recognition model. The recovery rate change result is 0 or 1, where 0 represents no change in the recovery rate and 1 represents a change in the recovery rate;

[0012] S4. Collect multi-source data of the shale oil reservoir to be predicted in the T time periods before the current moment, form a characteristic matrix of the shale oil reservoir to be predicted, and input the characteristic matrix of the shale oil reservoir to be predicted into the recovery rate change recognition model to obtain the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted;

[0013] S5. Fit the similarity between the characteristic matrix of the sample and the characteristic matrix of the shale oil reservoir to be predicted, obtain a similarity list of the shale oil reservoir to be predicted and the sample, find the sample with the largest similarity, and use the sample with the largest similarity as the reference sample of the shale oil reservoir to be predicted;

[0014] S6. Predict the recovery rate of the shale oil reservoir to be predicted according to the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted, and the reference sample of the shale oil reservoir to be predicted.

[0015] Furthermore, the geological parameters include porosity, permeability, and reservoir burial depth, the fluid property parameters include crude oil viscosity, density, and oil saturation, and the development process parameters include production method, number of fracturing operations, injected fluid volume, and production time.

[0016] Furthermore, characteristic parameters are extracted from porosity, permeability, and reservoir burial depth, including the mean porosity, mean permeability, and mean burial depth; characteristic parameters are extracted from crude oil viscosity, density, and oil saturation, including the mean rate of change of crude oil viscosity with pressure, the mean rate of change of density with pressure, and the mean oil saturation; characteristic parameters are extracted from production method, number of fracturing operations, injected fluid volume, and production time, including the one-hot encoding of the production method, the mean number of fracturing operations, the mean injection volume, and the mean injection time.

[0017] Furthermore, a characteristic matrix for each sample is constructed:

[0018]

[0019] where X j is the characteristic matrix of the j-th sample, and x j,1 , x j,2 , x j,3 , x j,4 , x j,5 , x j,6 , x j,7 , x j,8 , x j,9 , x j,10 are respectively the mean porosity, mean permeability, mean burial depth, mean rate of change of crude oil viscosity with pressure, mean rate of change of density with pressure, mean oil saturation, one-hot encoding, mean number of fracturing operations, mean injection volume, and mean injection time of the j-th sample, j is the index of the sample, m is the total number of samples, j is the index of the sample, m is the total number of samples;

[0020] A weight vector W is set, and each characteristic corresponds to a weight coefficient:

[0021]

[0022] where ω1, ω2, ω3, ω4, ω5, ω6, ω7, ω8, ω9, ω 10 are respectively the weight coefficients of the mean porosity, mean permeability, mean burial depth, mean rate of change of crude oil viscosity with pressure, mean rate of change of density with pressure, mean oil saturation, one-hot encoding, mean number of fracturing operations, mean injection volume, and mean injection time. On the basis of ω1 + ω2 + ω3 + ω4 + ω5 + ω6 + ω7 + ω8 + ω9 + ω 10 = 1, let 0 < ω7 < ω4 < ω3 < ω 10 < ω5 < ω9 < ω8 < ω1 < ω2 < ω6 < 1;

[0023] Weighted feature vector X of the sample j,wei :[[]]

[0024]

[0025] Furthermore, the feature matrix of the shale reservoir to be predicted:

[0026]

[0027] where X new is the feature matrix of the shale reservoir to be predicted, and x new,1 , x new,2 , x new,3 , x new,4 , x new,5 , x new,6 , x new,7 , x new,8 , x new,9 , x new,10 are respectively the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time of the shale reservoir to be predicted;

[0028] Weighted feature vector X of the shale reservoir to be predicted new,wei :[[]]

[0029]

[0030] Furthermore, obtain the similarity between the shale reservoir to be predicted and the sample. The specific formula is as follows:

[0031] Calculate the cosine similarity between the j-th sample and the shale reservoir to be predicted:

[0032]

[0033] where Sim(X new,eei , X j,wei ) is the cosine similarity between the j-th sample and the shale reservoir to be predicted, ||X new,wei || is the modulus of the weighted feature vector of the shale reservoir to be predicted, ||X j,wei || is the modulus of the weighted feature vector of the sample, ω β is the weight coefficient of the β-th feature, x new,β is the β-th feature of the shale reservoir to be predicted, x j,β is the β-th feature of the j-th sample, and β is the index of the feature, β ∈ [1, 10].

[0034] Further, arrange the similarities between the m samples and the shale oil reservoir to be predicted in descending order, and set a similarity threshold YZ. Based on this, select the reference samples for the shale oil reservoir to be predicted, with the following rules:

[0035] 1) If the reference sample is unique and the similarity is higher than YZ, then use the average recovery factor of this reference sample as the recovery factor of the shale oil reservoir to be predicted;

[0036] 2) If the reference samples are not unique and the similarity is higher than YZ;

[0037] If at least one of the reference samples has no change in the recovery factor, then use the average value of the recovery factors of the reference samples with no change in the recovery factor as the recovery factor of the shale oil reservoir to be predicted;

[0038] If all the reference samples have changes in the recovery factor, then use the average value of the recovery factors of all the reference samples as the recovery factor of the shale oil reservoir to be predicted;

[0039] 3) If the similarity of the reference sample is not higher than YZ, determine the recovery factor of the shale oil reservoir to be predicted based on the result of the change in the recovery factor;

[0040] If there is no change in the result of the change in the recovery factor, then the recovery factor of the shale oil reservoir to be predicted is the average recovery factor in the previous T time periods before the current moment;

[0041] If there is a change in the result of the change in the recovery factor and the reference sample with the highest similarity is unique, use the average recovery factor of the reference sample as the recovery factor of the shale oil reservoir to be predicted;

[0042] If there is a change in the result of the change in the recovery factor and the reference samples with the highest similarity are not unique, use the average value of the recovery factors of all the reference samples as the recovery factor of the shale oil reservoir to be predicted.

[0043] A device for predicting the recovery factor of a shale oil reservoir, which is used to execute any one of the above methods for predicting the recovery factor of a shale oil reservoir, includes:

[0044] A data acquisition module, which is used to collect multi-source data of samples in consecutive T historical time periods. The samples include shale oil reservoirs with no change in the recovery factor and shale oil reservoirs with changes in the recovery factor. Based on the shale oil reservoirs with no change in the recovery factor and shale oil reservoirs with changes in the recovery factor, calculate their average recovery factors in T historical time periods respectively. The multi-source data includes geological parameters, fluid property parameters, and development process parameters;

[0045] A feature matrix construction module, which is used to extract feature parameters from the collected geological parameters, fluid property parameters, and development process parameters respectively, and construct a feature matrix of each sample in consecutive T historical time periods based on the extracted feature parameters;

[0046] A change recognition model construction module is used to construct a recovery factor change recognition model. Taking the feature matrices of samples in T historical time periods as inputs, and using the recovery factor change results and the corresponding average recovery factors as labels to train the model, the recovery factor change recognition model is trained. The recovery factor change result is 0 or 1, where 0 represents no change in the recovery factor and 1 represents a change in the recovery factor.

[0047] A preliminary recovery factor acquisition module is used to collect multi-source data of the shale oil reservoir to be predicted in the T time periods before the current moment, form a feature matrix of the shale oil reservoir to be predicted, input the feature matrix of the shale oil reservoir to be predicted into the recovery factor change recognition model, and obtain the recovery factor change result and the corresponding average recovery factor of the shale oil reservoir to be predicted.

[0048] A fitting module is used to perform similarity fitting on the feature matrix of the sample and the feature matrix of the shale oil reservoir to be predicted, obtain a similarity list between the shale oil reservoir to be predicted and the sample, find the sample with the largest similarity, and use the sample with the largest similarity as the reference sample of the shale oil reservoir to be predicted.

[0049] A final recovery factor prediction module is used to predict the recovery factor of the shale oil reservoir to be predicted according to the recovery factor change result and the corresponding average recovery factor of the shale oil reservoir to be predicted, and the reference sample of the shale oil reservoir to be predicted.

[0050] A readable storage medium is used to store a computer program, and when the computer program is executed by a processor, it implements any one of the above-mentioned shale oil reservoir recovery factor prediction methods.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] The present invention widely collects multi-source data, covering conventional geological information and geological parameters such as the average porosity and average permeability, which intuitively reflects the ability of the rock to store and transmit fluids; collects the average change rate of crude oil viscosity, the average change rate of density and the average oil saturation, comprehensively considering the influence of fluid state on the recovery factor; collects one-hot encoding of the production method, average number of fracturing times, average water injection volume and average water injection time and other production process parameters, deeply excavating the role of different processes on the recovery factor. Based on these rich data, through in-depth analysis and feature extraction, when constructing the feature matrix, the average value is calculated from the geological parameters to reflect the long-term average characteristics of the oil reservoir, and key features are extracted from the production process parameters to scientifically evaluate the implementation effect of the process. This series of operations lays a solid foundation for subsequent model construction and analysis, making up for the deficiencies of the prior art in terms of one-sided data collection and shallow analysis.

[0053] In terms of model construction and prediction, thanks to the comprehensively and deeply mined data in the early stage, the established identification model for recovery factor changes can be trained based on the feature matrices of a large number of samples. Data of multiple shale oil reservoirs are collected over consecutive T historical time periods, covering different combinations of geological conditions, fluid properties, and production techniques, enabling the model to fully learn the complex relationships between various factor combinations and recovery factor changes, and draw rules from a large number of actual samples, so as to better handle the complex and changeable situations of actual oil reservoirs. Moreover, due to the continuous collection of multi-time period data, when the geological conditions, fluid properties, or production techniques of the oil reservoir change, new data can be fed back in a timely manner, and the model can accurately reflect the real situation through retraining or parameter adjustment. In addition, taking the recovery factor change result as a trend guidance and combining it with the similar case data of the reference samples, considering from two key perspectives, the change result gives the dynamic trend of the current oil reservoir recovery factor, while the reference samples provide data support in actual cases. The combination of the two avoids the one-sidedness of single-factor prediction and greatly improves the accuracy of prediction. Brief Description of the Drawings

[0054] Figure 1 It is a schematic flow chart of the overall method of the present invention;

[0055] Figure 2 It is a block diagram of the module composition of the present invention. Detailed Description of the Embodiments

[0056] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in combination with specific embodiments.

[0057] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second", and similar terms used in the present invention do not indicate any order, quantity, or importance, but are only used to distinguish different components. The terms such as "include" or "comprise" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connect" or "couple" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", "right", etc. are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0058] Example 1:

[0059] Please refer to Figure 1 , the present invention provides a technical solution:

[0060] A method for predicting the recovery rate of a shale oil reservoir, the specific steps including:

[0061] S1. Collect multi-source data of the sample in consecutive T historical time periods. The sample includes shale oil reservoirs with unchanged recovery rate and shale oil reservoirs with changed recovery rate. Based on the shale oil reservoirs with unchanged recovery rate and shale oil reservoirs with changed recovery rate, calculate their average recovery rates in T historical time periods respectively. The multi-source data includes geological parameters, fluid property parameters and development process parameters;

[0062] On the basis of the above embodiments, the geological parameters include porosity, permeability and reservoir burial depth, the fluid property parameters include crude oil viscosity, density and oil saturation, and the development process parameters include production method, number of fracturing times, injection fluid volume and production time.

[0063] Among them, in the consecutive T historical time periods, the production method is one of vertical well production, horizontal well production, fracturing production, water injection production and gas injection production.

[0064] S2. Extract characteristic parameters from the collected geological parameters, fluid property parameters and development process parameters respectively. Based on the extracted characteristic parameters, construct the characteristic matrix of each sample in consecutive T historical time periods;

[0065] On the basis of the above embodiments, extract characteristic parameters from porosity, permeability and reservoir burial depth, including average porosity, average permeability and average burial depth, extract characteristic parameters from crude oil viscosity, density and oil saturation, including average rate of change of crude oil viscosity with pressure, average rate of change of density with pressure and average oil saturation, and extract characteristic parameters from production method, number of fracturing times, injection fluid volume and production time, including one-hot encoding of production method, average number of fracturing times, average water injection volume and average water injection time.

[0066] On the basis of the above embodiments, collect multiple groups of data by setting multiple monitoring points in different shale oil reservoir areas. Each monitoring point includes geological parameters, fluid property parameters and development process parameters in T historical time periods.

[0067] On the basis of the above embodiments, set multiple monitoring points at different positions of the shale oil reservoir. These monitoring points are distributed in different areas and form a monitoring network within the entire shale oil reservoir range, so as to obtain geological parameters, fluid property parameters and development process parameters from various different positions.

[0068] For each monitoring point, there are geological parameters (porosity, permeability, reservoir burial depth), fluid property parameters (crude oil viscosity, density, oil saturation) and development process parameters (production method, number of fracturing times, injection fluid volume, production time) in T time periods.

[0069] Based on the above embodiments, characteristic parameters are extracted from the collected geological parameters, fluid property parameters, and development process parameters, including the mean porosity, mean permeability, mean burial depth, change rate of crude oil viscosity, change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time. The specific process is as follows:

[0070] Set the duration of each time period to 1 day, then the multi-source data in the i-th historical time period is S i , where i is the index of the historical time period, and i ∈ [1, T], and T is the total number of historical time periods;

[0071] The mean porosity is used to measure the average porosity level of the sample in consecutive T historical time periods. The mean porosity is calculated by the following formula:

[0072]

[0073] where, is the mean porosity, and φ i is the porosity in the i-th historical time period;

[0074] The mean permeability is used to measure the average permeability level of the sample in consecutive T historical time periods. The mean permeability is calculated by the following formula:

[0075]

[0076] where, is the mean permeability, and k i is the permeability in the i-th historical time period;

[0077] The mean burial depth is used to measure the average burial depth condition of the sample in consecutive T historical time periods. The mean burial depth is calculated by the following formula:

[0078]

[0079] where, is the mean burial depth, and H i is the burial depth in the i-th historical time period;

[0080] The mean change rate of crude oil viscosity is used to measure the average rate of change of crude oil viscosity with pressure in adjacent historical time periods. The mean change rate of crude oil viscosity is calculated by the following formula:

[0081]

[0082] where μ P is the mean change rate of crude oil viscosity, and μ i is the crude oil viscosity in the i-th historical time period, and μi-1 is the viscosity of crude oil in the (i - 1)-th historical time period, P i is the reservoir pressure corresponding to the i-th historical time period, P i-1 is the reservoir pressure corresponding to the (i - 1)-th historical time period;

[0083] The average density change rate is used to measure the average rate of density change with pressure of the sample in adjacent historical time periods. The average density change rate is calculated by the following formula:

[0084]

[0085] where ρ P is the average density change rate, ρ i is the density in the i-th historical time period, ρ i-1 is the density in the (i - 1)-th historical time period, P i is the reservoir pressure corresponding to the i-th historical time period, P i-1 is the reservoir pressure corresponding to the (i - 1)-th historical time period;

[0086] The average oil saturation is used to measure the average level of oil saturation of the sample in consecutive T historical time periods. The average oil saturation is calculated by the following formula:

[0087]

[0088] where is the average oil saturation, S O,i is the oil saturation in the i-th historical time period;

[0089] The one-hot encoding of the production method is set as follows:

[0090] The production methods of shale reservoirs include vertical well production, horizontal well production, fracturing production, water injection production, and gas injection production. When vertical well production is adopted, vertical well production is represented as [1, 0, 0, 0, 0] under one-hot encoding; when horizontal well production is adopted, horizontal well production is represented as [0, 1, 0, 0, 0] under one-hot encoding, and so on;

[0091] The average number of fracturing times is used to measure the average frequency of fracturing operations of the sample in consecutive T historical time periods. The average number of fracturing times is calculated by the following formula:

[0092]

[0093] where is the average number of fracturing times, N f,i is the number of fracturing times in the i-th historical time period;

[0094] The average water injection volume is used to measure the average level of the water volume injected into the reservoir by the sample in consecutive T historical time periods. The average water injection volume is calculated by the following formula:

[0095]

[0096] Wherein, is the average water injection volume, and V ω,i is the water injection volume in the i-th historical time period;

[0097] The average water injection time is used to measure the average duration of each water injection by the sample in consecutive T historical time periods. The average water injection time is calculated by the following formula:

[0098]

[0099] Wherein, is the average water injection time, and t ω,i is the water injection time in the i-th historical time period.

[0100] On the basis of the above embodiments, after collecting the average porosity, average permeability, average burial depth, average change rate of crude oil viscosity, average change rate of density, average oil saturation, average number of fracturing times, average water injection volume, and average water injection time, maximum-minimum normalization processing is performed on these parameters respectively, and then the normalized data is used for the subsequent analysis processing, so that various data can be analyzed and processed under the same dimension in the subsequent analysis processing.

[0101] S3. Construct a recognition model for the change in recovery factor. Use the feature matrix of the sample in T historical time periods as the input, and use the change result of the recovery factor and the corresponding average recovery factor as labels to train the model. Train the recognition model for the change in recovery factor. The change result of the recovery factor is 0 or 1, where 0 represents no change in the recovery factor, and 1 represents a change in the recovery factor;

[0102] On the basis of the above embodiments, construct the feature matrix of each sample:

[0103]

[0104] Wherein, X j is the feature matrix of the j-th sample, and x j,1 , x j,2 , x j,3 , x j,4 , x j,5 , x j,6 , x j,7 , x j,8 , x j,9 , x j,10They are the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time of the j-th sample, respectively. Here, j is the index of the sample, and m is the total number of samples;

[0105] Set the weight vector W, and each feature corresponds to a weight coefficient:

[0106]

[0107] Among them, ω1, ω2, ω3, ω4, ω5, ω6, ω7, ω8, ω9, ω 10 They are the weight coefficients of the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time, respectively;

[0108] The mean oil saturation directly reflects the enrichment degree of crude oil in the reservoir and is the key factor determining the recovery rate. A higher oil saturation means more crude oil can be exploited, which has the most direct and significant impact on the recovery rate. For example, under similar other conditions, the theoretical recovery rate of a reservoir with a high oil saturation will be higher. Therefore, ω6 is given the largest weight.

[0109] Permeability determines the flow ability of crude oil in the pores of reservoir rocks. Good permeability can make crude oil flow more smoothly towards the production well. Even when the oil saturation is certain, the improvement of permeability can effectively enhance the recovery rate. For example, in areas with high permeability, crude oil is more easily exploited. Compared with other factors such as porosity, its role in enhancing the recovery rate is more obvious. Therefore, the weight ω2 is relatively large.

[0110] Porosity provides the space for storing crude oil in the reservoir. Although a large porosity does not directly equal efficient exploitation of crude oil, it is the basic condition for crude oil storage. To a certain extent, porosity affects factors such as oil saturation and permeability, and has an important indirect impact on the recovery rate. Therefore, the weight ω1 ranks third.

[0111] Fracturing is an important means to improve the permeability of shale reservoirs. By increasing the number of fracturing times, more artificial fractures can be created to improve the seepage channels of crude oil, thus significantly affecting the recovery rate. During the development process, the transformation effect of fracturing operations on the reservoir is obvious. Therefore, the weight ω8 is relatively high.

[0112] In low-permeability shale oil reservoirs, the reservoir itself has extremely low permeability and high flow resistance for crude oil. At this time, fracturing operations become the core means to improve the reservoir's seepage capacity. By forming an artificial fracture network through fracturing, efficient flow channels can be provided for crude oil, and the effect on enhancing the recovery rate is immediate. In contrast, although an increase in the injection volume can also provide displacement power, if the reservoir permeability is not fundamentally improved, the injected water is prone to form an ineffective cycle and cannot effectively displace the crude oil. In such reservoirs, the number of fracturing operations has a more crucial impact on the recovery rate. Therefore, ω8 > ω9;

[0113] The injection volume plays a role in providing displacement energy during the water flooding process. An appropriate injection volume can effectively push the crude oil towards the production well and directly affect the recovery rate. Although the density change rate reflects the change in the physical properties of crude oil with pressure, its direct impact on the recovery rate is relatively small. For example, adjusting the injection volume can directly change the fluid flow state in the reservoir, while density changes mainly indirectly affect the recovery rate by influencing the flow resistance of crude oil. Therefore, ω9 > ω5.

[0114] The injection time affects the stability and sustainability of water injection. An appropriate injection time can ensure the full play of the water flooding effect and has a certain impact on the recovery rate. If the injection time is too short, an effective displacement system cannot be formed, and a large amount of crude oil cannot be driven to the bottom of the well; if the injection time is too long, it may lead to imbalance of formation pressure, causing problems such as water channeling and reducing the oil displacement efficiency. Burial depth, as an important geological property of shale oil reservoirs, has a less direct impact on the recovery rate in the short term. Burial depth mainly indirectly affects the recovery rate by influencing formation temperature, pressure, and rock physical properties. In contrast, the water injection operation can directly affect the water flooding process by adjusting the injection time in the short term. Therefore, considering the directness and short-term effects of their impacts on the recovery rate comprehensively, when constructing a parameter weight system for evaluating the recovery rate, the weight ω 10 assigned to the injection time is greater than the weight ω3 assigned to the burial depth, that is, ω 10 > ω3.

[0115] The change rate of crude oil viscosity reflects the change in the fluidity of crude oil with pressure, which has an important impact on the flow of crude oil in the reservoir and thus affects the recovery rate. Although the one-hot encoding of the production method is also related to the recovery rate, it only classifies different production methods and does not directly reflect the impact of changes in the physical properties of fluids in the reservoir on the recovery rate like the change rate of crude oil viscosity. Therefore, ω4 > ω7.

[0116] Therefore, on the basis of ω1 + ω2 + ω3 + ω4 + ω5 + ω6 + ω7 + ω8 + ω9 + ω 10 = 1, let 0 < ω7 < ω4 < ω3 < ω 10 < ω5 < ω9 < ω8 < ω1 < ω2 < ω6 < 1.

[0117] Weighted feature vector X of the sample j,wei :

[0118]

[0119] On the basis of the above embodiments, the recovery rate change recognition model is composed of a deep learning network based on a multi-layer perceptron. The deep neural network of the multi-layer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer, and an output layer. The first hidden layer, the second hidden layer, and the third hidden layer all have at least two neurons, and all use ReLU as the activation function;

[0120] In the recovery rate change recognition model, the input features of the deep learning network of the multi-layer perceptron include: the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time in the feature matrix of the sample, a total of 10 features.

[0121] The structure of the deep learning network of the multi-layer perceptron is as follows:

[0122] Input layer: Receives the input of 10 features;

[0123] First hidden layer: Has 128 neurons and uses ReLU as the activation function;

[0124] Second hidden layer: Has 64 neurons and also uses the ReLU activation function;

[0125] Third hidden layer: Has 32 neurons and uses the ReLU activation function;

[0126] Output layer: Has 1 neuron and outputs the recovery rate change result and the corresponding average recovery rate.

[0127] The process of training the recovery rate change recognition model is as follows:

[0128] Using the feature matrix of the sample in T historical time periods as the input quantity, and the recovery rate change result and the corresponding average recovery rate as the labels for training, and using the mean square error as the loss function. When the mean square error is within the range of [0, 0.01], the training of the recovery rate change recognition model is completed.

[0129] S4. Collect multi-source data of the shale oil reservoir to be predicted in the T time periods before the current moment, form the feature matrix of the shale oil reservoir to be predicted, input the feature matrix of the shale oil reservoir to be predicted into the recovery rate change recognition model, and obtain the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted;

[0130] Based on the above embodiments, the characteristic matrix of the shale oil reservoir to be predicted is as follows:

[0131]

[0132] Among them, X new is the characteristic matrix of the shale oil reservoir to be predicted, and x new,1 , x new,2 , x new,3 , x new,4 , x new,5 , x new,6 , x new,7 , x new,8 , x new,9 , x new,10 are respectively the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time of the shale oil reservoir to be predicted;

[0133] The weighted characteristic vector X new,wei of the shale oil reservoir to be predicted is as follows:

[0134]

[0135] S5. Fit the similarity between the characteristic matrix of the sample and the characteristic matrix of the shale oil reservoir to be predicted, obtain the similarity list between the shale oil reservoir to be predicted and the sample, find the sample with the largest similarity, and use the sample with the largest similarity as the reference sample for the shale oil reservoir to be predicted;

[0136] Based on the above embodiments, the similarity between the shale oil reservoir to be predicted and the sample is obtained, and the specific formula is as follows:

[0137] Calculate the cosine similarity between the j-th sample and the shale oil reservoir to be predicted:

[0138]

[0139]

[0140] Among them, Sim(X new,wei , X j,wei ) is the cosine similarity between the j-th sample and the shale oil reservoir to be predicted, ||X new,wei || is the modulus of the weighted characteristic vector of the shale oil reservoir to be predicted, ||X j,wei || is the modulus of the weighted characteristic vector of the sample, ω β is the weight coefficient of the β-th feature, x new,β is the β-th feature of the shale oil reservoir to be predicted, x j,β is the β-th feature of the j-th sample, and β is the index of the feature, β ∈ [1, 10].

[0141] S6. Predict the recovery factor of the shale oil reservoir to be predicted based on the change result of the recovery factor of the shale oil reservoir to be predicted and the corresponding average recovery factor, as well as the reference samples of the shale oil reservoir to be predicted.

[0142] Based on the above embodiments, the change result of the recovery factor predicted by the model can reflect the current change trend of the recovery factor of the shale oil reservoir to be predicted. This is a macroscopic judgment index, which provides a guiding direction for the prediction of the recovery factor. For example, if the result is 1, it indicates that there is a change in the recovery factor. Then, when making a prediction, it is necessary to consider the possible change factors, rather than simply judging according to the historical average situation.

[0143] The reference samples obtained by similarity are the historical samples that are most similar to the shale oil reservoir to be predicted in terms of geological parameters, fluid property parameters, development process parameters, etc. These reference samples contain various complex factors and results in the actual mining process, and can provide specific and comparable case references for the shale oil reservoir to be predicted.

[0144] Based on the above embodiments, predicting the recovery factor of the shale oil reservoir to be predicted by comprehensively considering the change result of the recovery factor and the reference samples has the following advantages compared with predicting the recovery factor of the shale oil reservoir according to the model output or based on the reference samples:

[0145] First, relying solely on the change result of the recovery factor predicted by the model may be too general and unable to take into account the specific characteristics and historical mining conditions of the actual oil reservoir. Relying solely on the reference samples may lead to prediction deviations due to some minor differences between the samples and the oil reservoir to be predicted. By combining the two, they can complement and verify each other, reduce the uncertainty brought by a single factor, and thus more accurately predict the recovery factor of the shale oil reservoir to be predicted.

[0146] Second, the recovery factor of the shale oil reservoir is affected by a variety of factors. Different oil reservoirs have differences in geological conditions, fluid properties, and development processes. Considering the information of these two aspects comprehensively can better adapt to the complexity of the shale oil reservoir, capture the key factors affecting the recovery factor and their interactions, and thus improve the accuracy and reliability of the prediction.

[0147] Based on the above embodiments, arrange the similarities of m samples with the shale oil reservoir to be predicted in descending order, and set a similarity threshold YZ. Accordingly, select the reference samples of the shale oil reservoir to be predicted, and the rules are as follows:

[0148] 1) If the reference sample is unique and the similarity is higher than YZ, then take the average recovery factor of this reference sample as the recovery factor of the shale oil reservoir to be predicted;

[0149] 2) If the reference samples are not unique and the similarity is higher than YZ;

[0150] If at least one reference sample with no change in recovery factor is included in the reference samples, the average recovery factor of the reference samples with no change in recovery factor is used as the recovery factor of the shale oil reservoir to be predicted.

[0151] If all the reference samples are reference samples with a change in recovery factor, the average recovery factor of all the reference samples is used as the recovery factor of the shale oil reservoir to be predicted.

[0152] 3) If the similarity of the reference samples is not higher than YZ, determine the recovery factor of the shale oil reservoir to be predicted according to the change result of the recovery factor.

[0153] If there is no change in the change result of the recovery factor, the recovery factor of the shale oil reservoir to be predicted is the average recovery factor in the previous T time periods before the current moment.

[0154] If there is a change in the change result of the recovery factor and the reference sample with the maximum similarity is unique, the average recovery factor of the reference sample is used as the recovery factor of the shale oil reservoir to be predicted.

[0155] If there is a change in the change result of the recovery factor and the reference samples with the maximum similarity are not unique, the average recovery factor of all the reference samples is used as the recovery factor of the shale oil reservoir to be predicted.

[0156] Among them, the average recovery factor mean refers to calculating the mean of the recovery factors in the historical time period and then calculating the mean of multiple reference samples.

[0157] Please refer to Figure 2 , and the present invention also provides a technical solution:

[0158] A device for predicting the recovery factor of a shale oil reservoir, the device is used to execute any one of the above-mentioned methods for predicting the recovery factor of a shale oil reservoir, and includes:

[0159] A data acquisition module, configured to acquire multi-source data of a sample in consecutive T historical time periods, the sample includes shale oil reservoirs with no change in recovery factor and with a change in recovery factor, and calculate their average recovery factors in T historical time periods respectively based on the shale oil reservoirs with no change in recovery factor and with a change in recovery factor, and the multi-source data includes geological parameter, fluid property parameter and development process parameter;

[0160] A feature matrix construction module, configured to extract feature parameters from the acquired geological parameters, fluid property parameters and development process parameters respectively, and construct a feature matrix of each sample in consecutive T historical time periods based on the extracted feature parameters;

[0161] A change recognition model construction module is used to construct a recovery rate change recognition model. Taking the feature matrices of samples in T historical time periods as inputs, and using the recovery rate change results and the corresponding average recovery rates as labels to train the model. The recovery rate change recognition model is trained, where the recovery rate change result is 0 or 1. 0 represents no change in the recovery rate, and 1 represents a change in the recovery rate.

[0162] A preliminary recovery rate acquisition module is used to collect multi-source data of the shale oil reservoir to be predicted in the T time periods before the current moment, form the feature matrix of the shale oil reservoir to be predicted, and input the feature matrix of the shale oil reservoir to be predicted into the recovery rate change recognition model to obtain the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted.

[0163] A fitting module is used to perform similarity fitting on the feature matrix of the sample and the feature matrix of the shale oil reservoir to be predicted, obtain a similarity list between the shale oil reservoir to be predicted and the sample, find the sample with the largest similarity, and use the sample with the largest similarity as the reference sample of the shale oil reservoir to be predicted.

[0164] A final recovery rate prediction module is used to predict the recovery rate of the shale oil reservoir to be predicted according to the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted, and the reference sample of the shale oil reservoir to be predicted.

[0165] A readable storage medium is used to store a computer program, and when the computer program is executed by a processor, it implements any one of the above-mentioned shale oil reservoir recovery rate prediction methods.

[0166] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0167] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by hardware, or a combination of computer software and hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0168] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0169] As described above, it is only the specific implementation manner of the present application. However, the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. A method for predicting the recovery factor of a shale oil reservoir, characterized in that: The specific steps include: S1. Collect multi-source data of the sample in consecutive T historical time periods. The sample includes shale oil reservoirs with unchanged and changed recovery rates. Based on the shale oil reservoirs with unchanged and changed recovery rates, calculate their average recovery rates in T historical time periods respectively. The multi-source data includes geological parameters, fluid property parameters, and development process parameters. S2. Extract characteristic parameters from the collected geological parameters, fluid property parameters, and development process parameters respectively. Based on the extracted characteristic parameters, construct a characteristic matrix of each sample in consecutive T historical time periods. S3. Construct a recovery rate change recognition model. Use the characteristic matrix of the sample in T historical time periods as the input, and use the recovery rate change result and the corresponding average recovery rate as labels to train the model. Train the recovery rate change recognition model. The recovery rate change result is 0 or 1, where 0 represents unchanged recovery rate and 1 represents changed recovery rate. S4. Collect multi-source data of the shale oil reservoir to be predicted in the T time periods before the current moment, form a characteristic matrix of the shale oil reservoir to be predicted, and input the characteristic matrix of the shale oil reservoir to be predicted into the recovery rate change recognition model to obtain the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted. S5. Fit the similarity between the characteristic matrix of the sample and the characteristic matrix of the shale oil reservoir to be predicted, obtain a similarity list between the shale oil reservoir to be predicted and the sample, find the sample with the largest similarity, and use the sample with the largest similarity as the reference sample of the shale oil reservoir to be predicted. S6. Predict the recovery rate of the shale oil reservoir to be predicted based on the recovery rate change result and the corresponding average recovery rate of the shale oil reservoir to be predicted, and the reference sample of the shale oil reservoir to be predicted.

2. The method for predicting the recovery factor of a shale oil reservoir according to claim 1, wherein: The geological parameters include porosity, permeability, and reservoir burial depth. The fluid property parameters include crude oil viscosity, density, and oil saturation. The development process parameters include production method, number of fracturing times, injected fluid volume, and production time.

3. The shale oil reservoir recovery factor prediction method according to claim 2, wherein: Extract characteristic parameters from porosity, permeability, and reservoir burial depth, including average porosity, average permeability, and average burial depth. Extract characteristic parameters from crude oil viscosity, density, and oil saturation, including average change rate of crude oil viscosity with pressure, average change rate of density with pressure, and average oil saturation. Extract characteristic parameters from production method, number of fracturing times, injected fluid volume, and production time, including one-hot encoding of production method, average number of fracturing times, average injection water volume, and average injection water time.

4. The shale oil reservoir recovery factor prediction method according to claim 3, characterized in that: Construct the characteristic matrix of each sample: Among them, X j is the feature matrix of the j-th sample, and x j,1 , x j,2 , x j,3 , x j,4 , x j,5 , x j,6 , x j,7 , x j,8 , x j,9 , x j,10 are respectively the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time of the j-th sample. j is the index of the sample, and m is the total number of samples; Set the weight vector W, and each characteristic corresponds to a weight coefficient: Among them, ω1, ω2, ω3, ω4, ω5, ω6, ω7, ω8, ω9, ω 10 are the weight coefficients of the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time, respectively. On the basis of ω1 + ω2 + ω3 + ω4 + ω5 + ω6 + ω7 + ω8 + ω9 + ω 10 = 1, let 0 < ω7 < ω4 < ω3 < ω 10 < ω5 < ω9 < ω8 < ω1 < ω2 < ω6 < 1; Weighted feature vector X of the sample j,wei :

5. The method for predicting the recovery factor of a shale oil reservoir according to claim 4, wherein: The characteristic matrix of the shale oil reservoir to be predicted: Among them, X new is the characteristic matrix of the shale oil reservoir to be predicted, and x new,1 , x new,2 , x new,3 , x new,4 , x new,5 , x new,6 , x new,7 , x new,8 , x new,9 , x new,10 are respectively the mean porosity, mean permeability, mean burial depth, mean change rate of crude oil viscosity, mean change rate of density, mean oil saturation, one-hot encoding, mean number of fracturing times, mean water injection volume, and mean water injection time of the shale oil reservoir to be predicted; Weighted feature vector X of the shale oil reservoir to be predicted new,wei :

6. The method for predicting the recovery factor of a shale oil reservoir according to claim 5, wherein: Obtain the similarity between the shale oil reservoir to be predicted and the sample. The specific formula is as follows: Calculate the cosine similarity between the j-th sample and the shale oil reservoir to be predicted: where Sim(X new,wei , X j,wei ) is the cosine similarity between the j-th sample and the shale oil reservoir to be predicted, ||X new,wei || is the norm of the weighted feature vector of the shale oil reservoir to be predicted, ||X j,wei || is the norm of the weighted feature vector of the sample, ω β is the weight coefficient of the β-th feature, x new,β is the β-th feature of the shale oil reservoir to be predicted, x j,β is the β-th feature of the j-th sample, and β is the index of the feature, β ∈ [1, 10].

7. The method for predicting the recovery factor of a shale oil reservoir according to claim 6, characterized in that: Arrange the similarities between the m samples and the shale oil reservoir to be predicted in descending order, and set a similarity threshold YZ. Accordingly, select the reference sample of the shale oil reservoir to be predicted. The rules are as follows: 1) If the reference sample is unique and the similarity is higher than YZ, then use the average recovery rate of this reference sample as the recovery rate of the shale oil reservoir to be predicted. 2) If the reference sample is not unique and the similarity is higher than YZ; If at least one reference sample with no change in recovery factor is included in the reference samples, the average recovery factor of the reference samples with no change in recovery factor is taken as the recovery factor of the shale oil reservoir to be predicted; If all the reference samples are reference samples with a change in recovery factor, the average recovery factor of all the reference samples is taken as the recovery factor of the shale oil reservoir to be predicted; 3) If the similarity of the reference samples is not higher than YZ, the recovery factor of the shale oil reservoir to be predicted is determined according to the result of the change in recovery factor; If there is no change in the result of the change in recovery factor, the recovery factor of the shale oil reservoir to be predicted is the average recovery factor in the previous T time periods before the current moment; If there is a change in the result of the change in recovery factor and the reference sample with the largest similarity is unique, the average recovery factor of the reference sample is taken as the recovery factor of the shale oil reservoir to be predicted; If there is a change in the result of the change in recovery factor and the reference samples with the largest similarity are not unique, the average recovery factor of all the reference samples is taken as the recovery factor of the shale oil reservoir to be predicted.

8. A shale oil reservoir recovery rate prediction device, which is used to execute the shale oil reservoir recovery rate prediction method according to any one of claims 1-7, and is characterized in that: Including: A data acquisition module, configured to acquire multi-source data of a sample in consecutive T historical time periods. The sample includes shale oil reservoirs with no change in recovery factor and shale oil reservoirs with a change in recovery factor. Based on the shale oil reservoirs with no change in recovery factor and shale oil reservoirs with a change in recovery factor, their average recovery factors in T historical time periods are calculated respectively. The multi-source data includes geological parameters, fluid property parameters, and development process parameters; A feature matrix construction module, configured to extract feature parameters from the acquired geological parameters, fluid property parameters, and development process parameters respectively, and construct a feature matrix of each sample in consecutive T historical time periods based on the extracted feature parameters; A change recognition model construction module, configured to construct a recovery factor change recognition model. The feature matrix of the sample in T historical time periods is used as the input, and the recovery factor change result and the corresponding average recovery factor are used as labels to train the model, and the recovery factor change recognition model is trained. The recovery factor change result is 0 or 1, where 0 represents no change in recovery factor and 1 represents a change in recovery factor; A preliminary recovery factor acquisition module, configured to acquire multi-source data of the shale oil reservoir to be predicted in the previous T time periods before the current moment, form a feature matrix of the shale oil reservoir to be predicted, input the feature matrix of the shale oil reservoir to be predicted into the recovery factor change recognition model, and obtain the recovery factor change result and the corresponding average recovery factor of the shale oil reservoir to be predicted; A fitting module, configured to perform similarity fitting on the feature matrix of the sample and the feature matrix of the shale oil reservoir to be predicted, obtain a similarity list of the shale oil reservoir to be predicted and the sample, find the sample with the largest similarity, and use the sample with the largest similarity as the reference sample of the shale oil reservoir to be predicted; A final recovery factor prediction module, configured to predict the recovery factor of the shale oil reservoir to be predicted according to the recovery factor change result and the corresponding average recovery factor of the shale oil reservoir to be predicted, and the reference sample of the shale oil reservoir to be predicted.

9. A readable storage medium, characterized in that: For storing a computer program, which when executed by a processor implements the method for predicting the recovery factor of a shale oil reservoir according to any one of claims 1-7.

Images