CN-122022024-A - Intelligent prediction method and device for production capacity of multi-layer oil well after pressure

CN122022024ACN 122022024 ACN122022024 ACN 122022024ACN-122022024-A

Abstract

The invention discloses an intelligent prediction method and device for capacity after multi-layer oil well production, which comprises the steps of calling a corresponding capacity prediction integration model according to geological units to which a target well belongs, wherein the capacity prediction integration model is trained according to the following modes that training data sets corresponding to the geological units are respectively obtained for different geological units in a target oil reservoir area, each training data set consists of historical capacity main control factors and historical capacity data of each reservoir section of each historical well in the geological unit, and the corresponding capacity prediction integration model is trained for each geological unit based on the training data sets. And inputting the input feature vector of each reservoir section of the target well into the corresponding productivity prediction integrated model, outputting the productivity predicted value corresponding to each reservoir section, wherein the input feature vector comprises the geological parameters and the fracturing construction parameters of the target well. The method can accurately and rapidly predict the productivity of each reservoir section of the multi-layer combined oil production well aiming at different geological units.

Inventors

ZHU RUIBIN
LI GENSHENG
LING FEI
SUN HAO
WANG LUYAO
XIE JINLIANG
YOU SHAOHUA
YAN ZHENGTING
Liao Qinzhuo
TIAN SHOUCENG

Assignees

中国石油大学(北京)

Dates

Publication Date: 20260512
Application Date: 20260119

Claims (10)

1. The intelligent prediction method for the production capacity of the multi-layer oil well after the production is pressed is characterized by comprising the following steps: The method comprises the steps of acquiring training data sets corresponding to different geological units in a target oil reservoir area according to the geological units to which a target well belongs, wherein the training data sets are composed of historical capacity main control factors and historical yield data of all reservoir sections of each historical well in the geological units, the historical capacity main control factors are obtained by screening historical geological parameters and historical fracturing construction parameters of all reservoir sections of the historical well, and the historical yield data are generated based on the historical geological parameters and the historical production data of all reservoir sections of the historical well; And inputting the input feature vector of each reservoir section of the target well into the corresponding productivity prediction integrated model, and outputting the productivity predicted value corresponding to each reservoir section, wherein the input feature vector comprises geological parameters and fracturing construction parameters of the target well.
2. The method according to claim 1, wherein the method further comprises: Acquiring historical data of each historical well in each geological unit in a target oil-bearing region, wherein the historical data comprises historical production data, historical geological parameters of each reservoir section and historical fracturing construction parameters, the historical production data at least comprises historical total yield, and the historical geological parameters at least comprise historical permeability and historical effective thickness; accordingly, the historical production data is generated based on historical geologic parameters and historical production data for each reservoir segment of the historical well, comprising: calculating, for each reservoir section of the historical well, a product of the historical permeability and the historical effective thickness of each reservoir section, and calculating a sum of the products of all reservoir sections; determining geological weights of all reservoir sections according to the ratio of the product of all reservoir sections to the sum of the products; splitting the historical total production into each reservoir segment based on the geological weight to generate historical production data independent of each reservoir segment.
3. The method of claim 1, wherein the historical energy production master factor is selected from historical geologic parameters and historical fracturing construction parameters for each reservoir interval of the historical well, comprising: evaluating historical geological parameters and historical fracturing construction parameters of each reservoir section of the historical well by adopting a plurality of different feature importance evaluation methods; calculating information entropy of each feature importance evaluation method according to the evaluation results of each parameter by each feature importance evaluation method, and carrying out weighted summation on each evaluation result based on the information entropy; And sorting the historical geological parameters and the historical fracturing construction parameters of each reservoir section of the historical well according to the weighted summation result, and selecting a preset number of parameters from the sorting result as a historical productivity main control factor.
4. The method of claim 3, wherein the plurality of different feature importance assessment methods comprises at least two of a gray correlation analysis, a maximum mutual information method, a SHAP interpretation method, a Pearson correlation analysis, a Stuffman correlation analysis.
5. The method of claim 1, wherein the capacity prediction integration model comprises a plurality of sub-models of different algorithm types, and wherein the training of the corresponding capacity prediction integration model for each geological unit based on the training dataset comprises: Based on the training data set, training and optimizing each sub-model in the productivity prediction integrated model corresponding to each geological unit by adopting a multi-round super-parameter optimization method.
6. The method of claim 5, wherein the plurality of sub-models of different algorithm types comprises at least two of a linear regression model, a tree model, and a gradient lifting model.
7. The method of claim 1, wherein inputting the input feature vector of each reservoir segment of the target well into the corresponding productivity prediction integration model, and outputting the productivity prediction value corresponding to each reservoir segment comprises: and inputting the input feature vectors of each reservoir section of the target well into the corresponding productivity prediction integration model, carrying out weighted average on the output of each sub-model in the corresponding productivity prediction integration model, and outputting the productivity prediction value corresponding to each reservoir section, wherein the weighted weight of each sub-model is determined based on the prediction performance index of each sub-model on the verification set.
8. The method according to claim 1, wherein the method further comprises: Calculating the capacity contribution duty ratio corresponding to each reservoir section based on the capacity predicted value corresponding to each reservoir section; determining potential primary and potential secondary intervals of the target well according to the capacity contribution ratio; Generating a differential fracturing design scheme for the target well according to the potential primary force interval and the potential secondary interval; and configuring a first fracturing construction parameter for the potential main force interval, and configuring a second fracturing construction parameter for the potential secondary interval, wherein the design strength of the first fracturing construction parameter is higher than that of the second fracturing construction parameter.
9. The utility model provides a multilayer closes behind oil recovery well pressure productivity intelligence prediction unit which characterized in that includes: The system comprises a training module, a capacity prediction integration model and a capacity prediction integration model, wherein the training module is used for respectively acquiring training data sets corresponding to different geological units in a target oil storage area, wherein the training data sets are composed of historical capacity main control factors and historical capacity data of all reservoir sections of each historical well in the geological units, the historical capacity main control factors are obtained by screening historical geological parameters and historical fracturing construction parameters of all reservoir sections of the historical well, and the historical capacity data are generated based on the historical geological parameters and the historical production data of all reservoir sections of the historical well; the calling module is used for calling the corresponding productivity prediction integration model according to the geological unit to which the target well belongs; And the prediction module is used for inputting the input feature vector of each reservoir section of the target well into the corresponding productivity prediction integrated model and outputting the productivity predicted value corresponding to each reservoir section, wherein the input feature vector comprises the geological parameters and the fracturing construction parameters of the target well.
10. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the steps of the method of any of claims 1 to 8.

Description

Intelligent prediction method and device for production capacity of multi-layer oil well after pressure Technical Field The invention relates to the technical field of oil and gas field development, in particular to an intelligent prediction method and device for the post-pressure productivity of a multi-layer oil production well. Background In oil and gas field development, hydraulic fracturing is a key technology for improving single well productivity, especially for unconventional reservoirs with low permeability, compactness and the like. The accurate prediction of the productivity after fracturing is a key premise of optimizing fracturing parameter design, formulating scientific exploitation strategies and evaluating exploitation benefits, directly influences the exploitation efficiency and economic benefits of multi-layer exploitation of low permeability oil reservoirs, and has an important supporting effect on pushing oil gas exploitation to 'intelligent, fine and efficient' transformation. However, the existing multi-layer production well post-pressure productivity prediction technology has obvious limitations that on one hand, the geological feature difference of different geological units is not considered, the unified model is adopted to predict the productivity of the whole area well, so that the suitability of the model and specific geological conditions is poor, the prediction accuracy is limited, and on the other hand, the existing model takes the total yield of the well as a prediction target, independent productivity prediction values of all reservoir sections cannot be accurately and rapidly output, and therefore the scientificity of layered fracturing optimization and development decision is restricted. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the specification provides an intelligent prediction method and device for capacity after multi-layer oil well pressure, which are used for solving the problems that the prediction adaptability in the prior art is poor, and the capacity predicted value of each reservoir section cannot be accurately and rapidly output. In a first aspect, an embodiment of the present disclosure provides an intelligent prediction method for post-pressure productivity of a multi-layer oil well, including: The method comprises the steps of acquiring training data sets corresponding to different geological units in a target oil reservoir area according to the geological units to which a target well belongs, wherein the training data sets are composed of historical capacity main control factors and historical yield data of all reservoir sections of each historical well in the geological units, the historical capacity main control factors are obtained by screening historical geological parameters and historical fracturing construction parameters of all reservoir sections of the historical well, and the historical yield data are generated based on the historical geological parameters and the historical production data of all reservoir sections of the historical well; And inputting the input feature vector of each reservoir section of the target well into the corresponding productivity prediction integrated model, and outputting the productivity predicted value corresponding to each reservoir section, wherein the input feature vector comprises geological parameters and fracturing construction parameters of the target well. In some embodiments, the method further comprises: Acquiring historical data of each historical well in each geological unit in a target oil-bearing region, wherein the historical data comprises historical production data, historical geological parameters of each reservoir section and historical fracturing construction parameters, the historical production data at least comprises historical total yield, and the historical geological parameters at least comprise historical permeability and historical effective thickness; accordingly, the historical production data is generated based on historical geologic parameters and historical production data for each reservoir segment of the historical well, comprising: calculating, for each reservoir section of the historical well, a product of the historical permeability and the historical effective thickness of each reservoir section, and calculating a sum of the products of all reservoir sections; determining geological weights of all reservoir sections according to the ratio of the product of all reservoir sections to the sum of the products; splitting the historical total production into each reservoir segment based on the geological weight to generate historical production data independent of each reservoir segment. In some embodiments, the historical energy production master factor is selected from historical geological parameters and historical fracturing construction parameters of each reservoir section of the historical well,