By adopting the method of the embodiment, 4 effective pressure early warnings are provided totally, which are consistent with the actually occurred drilling accidents. Therefore, the method in the embodiment has high accuracy of the pressure prediction result, and can prevent drilling accidents according to the pressure prediction result.

According to the pressure while drilling prediction method based on the comprehensive logging parameters, the first comprehensive logging parameter of the target well is obtained, and the formation pressure of the corresponding depth in the target well is predicted according to the first comprehensive logging parameter and the pressure prediction model. The pressure prediction model is obtained by combining DNN model training based on the pressure coefficient of the drilled well and the second comprehensive logging parameter, and the geological conditions of the drilled well and the target well are equivalent. Due to the characteristic of large data volume of the comprehensive logging parameters, the prediction accuracy of the pressure prediction model obtained by training according to the pressure coefficient of the drilled well and the second comprehensive logging parameters is higher; and because the geological conditions of the drilled well and the target well are equivalent, the pressure coefficient of the stratum at the corresponding depth in the target well can be accurately predicted according to the pressure prediction model and the first comprehensive logging parameter corresponding to the target well.

The above embodiments illustrate how a pressure prediction model may be used to make a pressure prediction for a target well. Next, how to train the pressure prediction model will be explained.

Fig. 2 is a schematic flow chart of a second embodiment of the pressure while drilling prediction method based on comprehensive logging parameters provided by the invention. As shown in fig. 2, the method of this embodiment may include:

s201, acquiring the pressure coefficient of the drilled well and a second comprehensive logging parameter.

Optionally, the pressure coefficient of the drilled well in the preset depth range and the second comprehensive logging parameter in the preset depth range may be obtained, and of course, the pressure coefficient of the drilled whole well section and the second comprehensive logging parameter of the whole well section may also be obtained.

Optionally, the second synthetic logging parameter may include: vertical well depth, hook height, while drilling, weight in suspension, weight on bit, casing pressure, drill plate rotational speed, first pump stroke, second pump stroke, drill plate torque, mud inlet density, mud inlet temperature, mud inlet conductivity, inlet resistivity, first mud pit volume, second mud pit volume, total pit volume, outlet flow, outlet density, outlet temperature, outlet resistivity, outlet conductivity, methane concentration, ethane concentration, propane concentration, isobutane concentration, n-butane concentration, isopentane concentration, total hydrocarbon concentration, DC index. It is understood that the second composite logging parameter may also include one or more of the above. Of course, the third comprehensive logging parameter may be other logging parameters, and is not limited to the above.

It should be noted that, under the action of the overburden, the compaction degree of the normally compacted stratum is correspondingly increased along with the increase of the buried depth, the rock density is relatively increased, the porosity of the rock is reduced, the drilling speed of the machine is reduced during drilling, and the drilling time is increased; when the drill meets a mudstone stratum in a pressure transition zone, the mechanical drilling speed is increased relative to the normal compaction of the mudstone during drilling due to the existence of the incompact mudstone, and the drilling time is reduced. In order to accurately reflect the relationship between the drilling time and the abnormal high-pressure layer, the influence of other factors on the drilling time must be eliminated. The DC index is a comprehensive index reflecting the formation drillability under the condition of eliminating the influence factors such as the bit pressure, the diameter of a drill bit, the rotating speed of a rotary table, the density of drilling fluid and the like. The method realizes that the drillability of all the layers to be drilled is compared under the same drilling condition, the abnormal section is researched and excavated, the abnormal high-pressure transition zone is found, and finally prediction and forecast are made. Under normal compaction, the DC index increases with increasing well depth. When drilling a transition zone across an abnormally high pressure formation, the DC index deviates from the normal compaction trend line in a decreasing direction. From which the position of the top of the transition zone can be predicted and abnormally high pressures predicted. The DC index is an extremely important parameter in the comprehensive logging parameters.

S202, preprocessing the second comprehensive logging parameter to obtain a preprocessed second comprehensive logging parameter.

And preprocessing the second comprehensive logging parameter for removing invalid data in the second comprehensive logging parameter.

In this embodiment, the second comprehensive logging parameter of the drilled well within the preset depth range is taken as an example for explanation.

In one possible implementation, the second composite logging parameter is pre-processed by:

and respectively carrying out statistical analysis on the acquired numerical values of each logging parameter of the drilled well within the preset depth range, determining the reasonable minimum value and the reasonable maximum value of each logging parameter, and removing abnormal values in each logging parameter. For example, Statistical analysis may be performed using Statistical Product and Service Solutions (SPSS) software tools. Invalid data in the second comprehensive logging parameter can be removed in a manual processing mode.

And performing the preprocessing on the second comprehensive logging parameters which are drilled, removing invalid data, and reducing the error of the pressure prediction result by ensuring the accuracy of the sample data for establishing the pressure prediction model so as to further improve the accuracy of the pressure prediction result of the pressure prediction model.

And S203, performing dimensionality reduction treatment by adopting a principal component analysis method according to the preprocessed second comprehensive logging parameter to obtain a third comprehensive logging parameter.

In the comprehensive logging parameters obtained in the actual drilling and exploitation process, a plurality of logging parameters have a correlation relationship, and when two logging parameters have a certain correlation relationship, the information reflected by the two logging parameters can be considered to have a certain overlap. Therefore, the logging parameters reflecting the same information are removed by a principal component analysis method, the logging parameters as few as possible are established, and the logging parameters are irrelevant in pairs but can reflect the information of the original logging parameters.

Principal Component Analysis (PCA) is a mathematical transformation method that can transform multiple indices into a few comprehensive indices. In this embodiment, the principal component analysis of the comprehensive logging parameters may be implemented by using simulation software such as MATLAB, where the process of implementing the principal component analysis by using simulation software such as MATLAB is similar to that in the prior art, and is not described herein again.

And performing principal component analysis on the preprocessed second comprehensive logging parameter to obtain a third comprehensive logging parameter. Optionally, the third composite logging parameter may include one or more of: weight on bit, time on bit, carousel speed, outlet density, inlet density, outlet conductance, inlet conductance. Of course, it is understood that the principal component analysis is performed on different drilled comprehensive logging parameters, and the obtained principal components may be different due to different drilled geological conditions, reservoir conditions, and the like, that is, the third comprehensive logging parameter may also be other logging parameters, and is not limited to the above-described logging parameters.

By analyzing the principal components of the comprehensive logging parameters, the data quantity participating in calculation can be reduced, and the calculation efficiency is improved.

And S204, establishing a pressure prediction model by adopting a DNN model according to the drilled third comprehensive logging parameters and the pressure coefficient. On the basis of the above embodiment, the pressure coefficient and the third comprehensive logging parameter may be data corresponding to a full well section of a drilled well; or the pressure coefficient and the third comprehensive logging parameter are the pressure coefficient and the third comprehensive logging parameter of the drilled well in a preset depth range. The following are described separately:

in one possible implementation, the pressure coefficient and the third comprehensive logging parameter are data corresponding to a full well section of the drilled well.

And taking the third comprehensive logging parameter and the pressure coefficient as training samples, training the DNN model, and stopping sample training when the output accuracy of the DNN model meets a preset condition to obtain a pressure prediction model.

Specifically, sample data is obtained according to the third comprehensive logging parameter and the pressure coefficient of the drilled whole well section at preset intervals and is used as a training sample of the DNN model, and the DNN model is trained. For example, the pressure coefficient corresponding to the drilled well, the bit pressure, the drilling time, the rotating speed of a rotary table, the outlet density, the inlet density, the outlet conductance and the inlet conductance are obtained at intervals of 0.125 m, multiple groups of obtained sample data are used as training samples of the DNN model, the DNN model is subjected to sample training, the linear relation coefficients among all layers of the DNN model are continuously optimized in the sample training process, when the output accuracy of the DNN model meets preset conditions, the sample training is stopped, and the linear relation coefficients among all layers of the DNN model are stored, so that the pressure prediction model is obtained.

In another possible embodiment, the pressure coefficient and the third comprehensive logging parameter are the pressure coefficient of the drilled well within a preset depth range and the third comprehensive logging parameter within the preset depth range.

Optionally, before the pressure prediction model is established by using the DNN model according to the third comprehensive logging parameter and the pressure coefficient, the comprehensive logging parameter and the pressure coefficient of the full well section corresponding to the drilled well are obtained by using a Back Propagation (BP) algorithm, the third comprehensive logging parameter and the pressure coefficient.

Specifically, a BP network is constructed through a BP algorithm, and the third comprehensive logging parameter of the drilled whole well section and the pressure coefficient of the drilled whole well section are obtained by combining the third comprehensive logging parameter of the drilled whole well section in the preset depth range and the pressure coefficient of the drilled whole well section in the preset depth range.

And further, training the DNN model by taking the third comprehensive logging parameter and the pressure coefficient of the whole well section as training samples to obtain a pressure prediction model.

The third comprehensive logging parameter and the pressure coefficient of the whole well section of the drilled well can be obtained according to the third comprehensive logging parameter and the pressure coefficient within the preset depth range of the drilled well through a BP algorithm, so that the sample data size is enlarged, and the accuracy of the pressure prediction result of the pressure prediction model is improved through a large amount of sample training of a DNN model.

It is understood that, with the method in this embodiment, a DNN model may also be used to establish a pressure prediction model corresponding to different wellbore intervals according to the third comprehensive logging parameter and the pressure coefficient of the different wellbore intervals that have been drilled. In the practical application process, different pressure prediction models are adopted for different well sections of the target well, so that a more accurate pressure prediction result can be obtained.

In this embodiment, the second comprehensive logging parameter is preprocessed by obtaining a pressure coefficient of the drilled well and the second comprehensive logging parameter, so as to obtain a preprocessed second comprehensive logging parameter, further, the PCA is used for performing dimensionality reduction processing according to the preprocessed second comprehensive logging parameter, so as to obtain a third comprehensive logging parameter, and a DNN model is used for establishing a pressure prediction model according to the drilled third comprehensive logging parameter and the pressure coefficient. By adopting a principal component analysis method to extract principal components capable of reflecting the second comprehensive parameter information, the data volume participating in calculation is reduced, and the calculation efficiency is improved; and (3) taking the drilled third comprehensive logging parameter and the pressure coefficient as sample data, and carrying out a large number of sample training on the DNN model to obtain the pressure prediction model with high pressure prediction result accuracy. Therefore, the pressure coefficient of the stratum at the corresponding depth in the target well can be accurately predicted by using the pressure prediction model and the first comprehensive logging parameter corresponding to the target well.

In the above embodiment, when the output accuracy of the DNN model meets the preset condition, the sample training is stopped to obtain the pressure prediction model, which can be implemented in the following manner:

in a possible implementation manner, during the sample training process of the DNN model, an error coefficient of the DNN model in each sample training is obtained. And when the sum of the error coefficients of the multiple sample training is smaller than a preset threshold value, considering that the output accuracy of the DNN model meets a preset condition, and stopping the sample training, thereby obtaining the pressure prediction model.

It should be noted that when the sum of the error coefficients of multiple sample training is smaller than the preset threshold, the DNN model is considered to be fully learned, and the error of the pressure prediction result is smaller, so that the accuracy of the pressure prediction result of the pressure prediction model is ensured to be higher.

In another possible implementation manner, the DNN model sample training times are obtained in the process of performing sample training on the DNN model. And when the sample training times are larger than the preset sample training times, considering that the output accuracy of the DNN model meets the preset condition, stopping the sample training, and obtaining the pressure prediction model.

It should be noted that when the DNN model sample training times are greater than the preset sample training times, the DNN model is considered to perform sufficient sample learning, so that it is ensured that the accuracy of the pressure prediction result of the pressure prediction model is high.

The pressure coefficient in the preset depth range of the drilled well and the comprehensive logging parameter in the preset depth range, the pressure coefficient value of the well section outside the preset depth range and the comprehensive logging parameter value of the well section outside the preset depth range are obtained in advance through a BP algorithm. There may be an error with the actually measured pressure coefficient of the drilled well, and therefore, after the pressure prediction model is established, the pressure prediction model may be calibrated according to the actually measured pressure coefficient of the well section outside the preset depth range of the drilled well and the third comprehensive logging parameter.

In some embodiments, when the output accuracy of the DNN model meets a preset condition, stopping the sample training, and obtaining the pressure prediction model, the method may further include the following steps:

and calibrating the pressure prediction model according to the actually measured pressure coefficient of the drilled well section outside the preset depth range and the third comprehensive logging parameter. In one possible implementation manner, the actually measured pressure coefficients of other well sections outside the preset depth range are compared with the pressure coefficient predicted by the pressure prediction model, the well section depth with the difference value between the predicted pressure coefficient and the actually measured pressure coefficient larger than the preset error value is obtained, the actually measured pressure coefficient corresponding to the well section depth and the third comprehensive logging parameter corresponding to the well section depth are input into the DNN model for sample learning again, and the pressure prediction model is calibrated. In other words, sample learning is performed again through the DNN model, and linear relation coefficients among layers of the DNN model are optimized to obtain the pressure prediction model with higher pressure prediction result accuracy. Particularly, for the stratum with a complex pressure system, the stratum pressure coefficient can be predicted more accurately, and the possibility of drilling accidents is reduced.

It should be noted that, in practical applications, the first comprehensive logging parameter of the target well is consistent with the selected third comprehensive logging parameter of the drilled well. For example, the drilled third composite logging parameter includes: weight on bit, time on bit, rate of the rotary table, density at the outlet, density at the inlet, conductance at the outlet, conductance at the inlet. Accordingly, the first integrated logging parameter for the target well comprises: weight on bit, time on bit, rate of the rotary table, density at the outlet, density at the inlet, conductance at the outlet, conductance at the inlet.

In the actual drilling and exploitation process, the comprehensive logging instrument can acquire a third comprehensive logging parameter of the target well in real time, the third comprehensive logging parameter is input into the pressure prediction model, and the pressure coefficient of the stratum at the corresponding depth is predicted, so that the real-time and accurate monitoring of the stratum pressure of the target well is realized.

Fig. 3 is a schematic structural diagram of a first embodiment of the pressure while drilling prediction apparatus based on comprehensive logging parameters provided by the present invention. As shown in fig. 3, theapparatus 30 includes: afirst acquisition module 31 and apressure prediction module 32.

The first obtainingmodule 31 is configured to obtain a first comprehensive logging parameter of a target well;

and thepressure prediction module 32 is used for predicting the pressure coefficient of the stratum at the corresponding depth in the target well according to the first comprehensive logging parameter and the pressure prediction model, wherein the pressure prediction model is obtained by combining the training of the DNN model based on the pressure coefficient of the drilled well and the second comprehensive logging parameter, and the geological conditions of the drilled well and the target well are equivalent.

The apparatus of this embodiment may be used to implement the technical solution of the method embodiment shown in fig. 1, and the implementation principle and the technical effect are similar, which are not described herein again.

Fig. 4 is a schematic structural diagram of a second embodiment of the pressure while drilling prediction apparatus based on comprehensive logging parameters provided by the present invention. As shown in fig. 4, the apparatus in this embodiment is based on the embodiment shown in fig. 3, and theapparatus 40 further includes: a second obtainingmodule 41, adata preprocessing module 42, a principalcomponent analyzing module 43, and abuilding module 44.

The second obtainingmodule 41 is configured to obtain a pressure coefficient of the drilled well and a second comprehensive logging parameter.

Adata preprocessing module 42, configured to preprocess the second comprehensive logging parameter to obtain a preprocessed second comprehensive logging parameter, where the preprocessing is used to remove invalid data in the second comprehensive logging parameter

And the principalcomponent analysis module 43 is configured to perform, according to the preprocessed second comprehensive logging parameter, dimensionality reduction processing by using PCA, and obtain a third comprehensive logging parameter.

Abuilding module 44, configured to build a pressure prediction model by using the DNN model according to the third comprehensive logging parameter and the pressure coefficient

The establishingmodule 44 is specifically configured to train the DNN model by using the third comprehensive logging parameter and the pressure coefficient as training samples, and when the output accuracy of the DNN model meets a preset condition, stop the sample training to obtain the pressure prediction model.

When the third comprehensive logging parameter and the pressure coefficient are the comprehensive logging parameter and the pressure coefficient of the drilled well in the preset depth range, before the DNN model is used to build the pressure prediction model according to the third comprehensive logging parameter and the pressure coefficient, thebuilding module 44 is further configured to:

and obtaining the comprehensive logging parameters and the pressure coefficients of the whole well section corresponding to the drilled well according to the preset BP algorithm, the third comprehensive logging parameters and the pressure coefficients.

Further, the establishingmodule 44 takes the third comprehensive logging parameter and the pressure coefficient of the whole well section as a training sample, trains the DNN model, and stops the sample training when the output accuracy of the DNN model meets a preset condition to obtain the pressure prediction model.

Optionally, the second composite logging parameters may include one or more of the following parameters: vertical well depth, hook height, hook load, while drilling, weight in suspension, weight in bit, vertical pressure, casing pressure, drill plate rotational speed, first pump stroke, second pump stroke, drill plate torque, mud inlet density, mud inlet temperature, mud inlet conductivity, inlet resistance, first mud pit volume, second mud pit volume, total pit volume, inlet flow, outlet flow, inlet density, outlet density, inlet temperature, outlet resistance, outlet conductance, methane concentration, ethane concentration, propane concentration, isobutane concentration, n-butane concentration, isopentane concentration, total hydrocarbon concentration, DC index.

Optionally, the first synthetic logging parameter may include one or more of the following parameters: weight on bit, time on bit, carousel speed, outlet density, inlet density, outlet conductance, inlet conductance.

In practical applications, the first integrated logging parameter of the target well is consistent with the selected third integrated logging parameter of the drilled well. For example, the drilled third composite logging parameter includes: weight on bit, time on bit, carousel speed, outlet density, inlet density, outlet conductance, inlet conductance. Accordingly, the first integrated logging parameter for the target well comprises: weight on bit, time on bit, carousel speed, outlet density, inlet density, outlet conductance, inlet conductance.

The apparatus of this embodiment may be used to implement the technical solution of the method embodiment shown in fig. 2, and the implementation principle and the technical effect are similar, which are not described herein again.

The pressure coefficient in the preset depth range of the drilled well and the comprehensive logging parameter in the preset depth range, the pressure coefficient value of the well section outside the preset depth range and the comprehensive logging parameter value of the well section outside the preset depth range are obtained through a back propagation algorithm in advance. There may be an error with the actually measured pressure coefficient of the drilled well, and therefore, after the pressure prediction model is established, the pressure prediction model may be calibrated according to the actually measured pressure coefficient of the drilled well section outside the preset depth range and the comprehensive logging parameters.

Optionally, the apparatus of the foregoing embodiment may further include: acalibration module 45.

Thecalibration module 45 is specifically configured to calibrate the pressure prediction model according to the actually measured pressure coefficient of the drilled well section outside the preset depth range and the third comprehensive logging parameter.

Fig. 5 is a schematic structural diagram of a third embodiment of the pressure while drilling prediction apparatus based on comprehensive logging parameters provided by the present invention. As shown in fig. 5, theapparatus 50 of the present embodiment includes: amemory 51 and aprocessor 52.

Thememory 51 may be a separate physical unit, and may be connected to theprocessor 52 via abus 53. Thememory 51 and theprocessor 52 may also be integrated, implemented by hardware, etc.

Thememory 51 is used for storing a program implementing the above method embodiment, which is called by theprocessor 52 to perform the operations of the above method embodiment.

Alternatively, when part or all of the methods of the above embodiments are implemented by software, theelectronic device 50 may only include a processor. The memory for storing the program is located outside theelectronic device 50, and the processor is connected to the memory by a circuit/wire for reading and executing the program stored in the memory.

Processor 52 may be a Central Processing Unit (CPU), a Network Processor (NP), or a combination of a CPU and an NP.

Theprocessor 52 may further include a hardware chip. The hardware chip may be an Application-Specific Integrated Circuit (ASIC), a Programmable Logic Device (PLD), or a combination thereof. The PLD may be a Complex Programmable Logic Device (CPLD), a Field-Programmable gate Array (FPGA), General Array Logic (GAL), or any combination thereof.

TheMemory 51 may include a Volatile Memory (Volatile Memory), such as a Random-Access Memory (RAM); the Memory may also include a Non-volatile Memory (Non-volatile Memory), such as a Flash Memory (Flash Memory), a Hard Disk Drive (HDD) or a Solid-state Drive (SSD); the memory may also comprise a combination of memories of the kind described above.

The apparatus of this embodiment may be used to implement the technical solutions of the method embodiments shown in fig. 1 and fig. 2, and the implementation principles and technical effects are similar, which are not described herein again.

The present invention also provides a program product, e.g., a computer storage medium, comprising: program for performing the above method when executed by a processor.

Those of ordinary skill in the art will understand that: all or a portion of the steps of implementing the above-described method embodiments may be performed by hardware associated with program instructions. The program may be stored in a computer-readable storage medium. When executed, the program performs steps comprising the method embodiments described above; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A pressure while drilling prediction method based on comprehensive logging parameters is characterized by comprising the following steps:

acquiring a first comprehensive logging parameter of a target well;

2. The method of claim 1, wherein prior to predicting the pressure coefficient of the formation at the corresponding depth in the target well based on the first integrated logging parameter and the pressure prediction model, further comprising:

obtaining the pressure prediction model according to the following steps:

3. The method of claim 2, wherein the building the pressure prediction model using the DNN model based on the third synthetic logging parameter and the pressure coefficient comprises:

4. The method of claim 2, wherein the third integrated logging parameter and the pressure coefficient are integrated logging parameters and pressure coefficients, respectively, within a predetermined depth range in the drilled well, and wherein before the building the pressure prediction model using the DNN model based on the third integrated logging parameter and the pressure coefficients, further comprises:

5. The method according to claim 3 or 4, wherein when the output accuracy of the DNN model meets a preset condition, stopping sample training, and after obtaining the stress prediction model, further comprising:

6. The method of any of claims 1 to 4, wherein the second synthetic logging parameters comprise one or more of the following parameters:

7. The method of any of claims 1 to 4, wherein the first synthetic logging parameter comprises one or more of:

8. A pressure while drilling prediction device based on comprehensive logging parameters is characterized by comprising:

9. A pressure while drilling prediction device based on comprehensive logging parameters is characterized by comprising: a processor and a memory;

the memory stores program instructions;

the processor executes the program instructions to perform the method of any one of claims 1 to 7.

10. A storage medium, comprising: carrying out a procedure;

the program is for performing the method of any one of claims 1 to 7 when executed by a processor.