Stand pipe pressure prediction method and device

Abstract

According to the riser pressure prediction method and device provided by the invention, the riser pressure prediction method based on the physical model and the riser pressure prediction method based on machine learning modeling are established, and the hybrid model is established by comprehensively utilizing the advantages of the two methods, so that riser pressure prediction under various different conditions is realized; and the occurrence of an abnormal condition is pre-judged in advance.

Description

Technical Field

[0001] The invention relates to the technical field of oil drilling, and in particular to a method and a device for predicting standpipe pressure. Background Art

[0002] During the drilling construction of oil resources, once the riser pressure becomes abnormal, it may cause serious and complex situations such as drilling tool leakage, broken drilling tools, water eye blockage, drill bit mud packing, overflow and well leakage, which may lead to damage to the on-site drilling tools and abandonment of oil and gas wells, and cause a series of environmental pollution problems.

[0003] Although the prediction of the riser pressure value can detect the abnormality of the riser pressure in advance and there are many complex situations that can be predicted, the logging data will only change when an abnormal situation occurs. Therefore, it is impossible to predict the occurrence of abnormal situations in advance through logging data. Summary of the invention

[0004] The present invention provides a method and device for predicting standpipe pressure, which realizes automatic data collection and standpipe pressure prediction by establishing a set of standpipe pressure prediction methods applicable to various situations.

[0005] In a first aspect, the present invention provides a method for predicting riser pressure, comprising:

[0006] Obtain real-time status data of the wellhead to be predicted;

[0007] Determining a target model from a pre-set physical model, a pre-trained riser pressure prediction neural network, and a pre-trained rig status prediction neural network according to the real-time status data;

[0008] If the target model is the physical model, based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is determined through the physical model;

[0009] If the target model is the standpipe pressure prediction neural network, the real-time status data is input into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted;

[0010] If the target model is the drilling rig state prediction neural network, the real-time state data is input into the drilling rig state prediction neural network to obtain a predicted kinetic energy coefficient;

[0011] Replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model, and passing the hybrid physical model;

[0012] Based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model.

[0013] Optionally, the training step of the standpipe pressure prediction neural network includes:

[0014] Acquire a riser pressure prediction training sample; the riser pressure training sample includes: historical state data and historical riser pressure at corresponding time; the historical state data includes: historical seismic data, historical drilling data and historical logging data;

[0015] The initially established prediction neural network is trained according to all the historical seismic data, the historical drilling data, the historical logging data and the historical standpipe pressure to obtain the standpipe pressure prediction neural network.

[0016] Optionally, the training step of the drilling rig prediction neural network includes:

[0017] Determine the first riser pressure prediction result corresponding to all the historical state data through a preset physical model;

[0018] Determining the second riser pressure prediction results corresponding to all the historical state data through a preset riser pressure prediction neural network;

[0019] The initially established drilling rig state detection model is trained according to the historical drilling rig state data, the first riser pressure prediction result and the second riser pressure prediction result to obtain a drilling rig state prediction neural network.

[0020] Optionally, based on the real-time status data, obtaining the riser pressure prediction data of the wellhead to be predicted through the hybrid physical model includes:

[0021] Acquire real-time status data of multiple adjacent wells, and determine the riser pressure prediction data of all adjacent wells through the hybrid physical model;

[0022] The hybrid physical model is curve-fitted using all the predicted data of the riser pressures of the adjacent wells to obtain the predicted data of the riser pressure of the wellhead to be predicted.

[0023] In a second aspect, the present invention provides a standpipe pressure prediction device, comprising:

[0024] An acquisition module is used to acquire real-time status data of the wellhead to be predicted;

[0025] A target model determination module, used to determine the target model from a preset physical model, a pre-trained standpipe pressure prediction neural network and a pre-trained drilling rig state prediction neural network according to the real-time status data;

[0026] A first prediction module, configured to determine the riser pressure prediction data of the wellhead to be predicted based on the real-time status data and through the physical model if the target model is the physical model;

[0027] A second prediction module is used for inputting the real-time status data into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted if the target model is the standpipe pressure prediction neural network;

[0028] a third prediction module, for inputting the real-time status data into the drilling rig status prediction neural network to obtain a predicted kinetic energy coefficient if the target model is the drilling rig status prediction neural network; and

[0029] Replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model; and,

[0030] Based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model.

[0031] Optionally, it also includes:

[0032] A sample data acquisition module is used to acquire a riser pressure prediction training sample; the riser pressure training sample includes: historical state data and historical riser pressure at corresponding time; the historical state data includes: historical seismic data, historical drilling data and historical logging data;

[0033] The target neural network training module is used to train the initially established prediction neural network based on all the historical seismic data, the historical drilling data, the historical logging data and the historical standpipe pressure to obtain the standpipe pressure prediction neural network.

[0034] Optionally, it also includes:

[0035] A first prediction result determination module, used to determine the first riser pressure prediction result corresponding to all the historical state data through a preset physical model;

[0036] A second prediction result determination module, used to determine the second riser pressure prediction results corresponding to all the historical state data through a preset riser pressure prediction neural network;

[0037] The drilling rig state prediction neural network generation module is used to train the initially established drilling rig state detection model according to the historical drilling rig state data, the first riser pressure prediction result and the second riser pressure prediction result to obtain the drilling rig state prediction neural network.

[0038] Optionally, the third prediction module includes:

[0039] A submodule for determining the prediction data of the riser pressure of adjacent wells, used for acquiring the real-time status data of a plurality of adjacent wells, and determining the prediction data of the riser pressure of all the adjacent wells through the hybrid physical model;

[0040] The curve fitting submodule is used to perform curve fitting on the hybrid physical model using all the prediction data of the riser pressures of the adjacent wells to obtain the prediction data of the riser pressure of the wellhead to be predicted.

[0041] In a third aspect, the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are performed.

[0042] In a fourth aspect, the present application provides a storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps in the method provided in the first aspect are performed.

[0043] It can be seen from the above technical solutions that the present invention has the following advantages:

[0044] The present invention provides a method and device for predicting standpipe pressure, the method comprising: acquiring real-time status data of a wellhead to be predicted; determining a target model from a preset physical model, a pre-trained standpipe pressure prediction neural network and a pre-trained drilling rig status prediction neural network according to the real-time status data; if the target model is the physical model, determining the standpipe pressure prediction data of the wellhead to be predicted through the physical model based on the real-time status data; if the target model is the standpipe pressure prediction neural network, inputting the real-time status data into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted; if the target model is the drilling rig status prediction neural network, inputting the real-time status data into the standpipe pressure prediction neural network to obtain the predicted kinetic energy coefficient; replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model, and through the hybrid physical model; based on the real-time status data, through the hybrid physical model, obtaining the standpipe pressure prediction data of the wellhead to be predicted. By establishing a riser pressure prediction method based on a physical model and a riser pressure prediction method based on machine learning modeling, and comprehensively utilizing the advantages of the two methods to establish a hybrid model, the riser pressure prediction under a variety of different situations can be achieved, thereby predicting the occurrence of abnormal situations in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1 This is a flow chart of a first embodiment of a method for predicting standpipe pressure according to the present invention;

[0047] Figure 2 This is a flow chart of a second embodiment of a method for predicting standpipe pressure according to the present invention;

[0048] Figure 3 The present invention is a structural block diagram of an embodiment of a device for optimizing the loading process of a bulk cargo terminal according to the present invention. DETAILED DESCRIPTION

[0049] The embodiment of the present invention provides a method and device for predicting riser pressure, which realizes automatic data collection and riser pressure prediction by establishing a set of methods for predicting riser pressure applicable to various situations.

[0050] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] For example, see Figure 1 , Figure 1 The present invention is a flow chart of a method for predicting standpipe pressure in Embodiment 1, comprising:

[0052] Step S101, obtaining real-time status data of the wellhead to be predicted;

[0053] Step S102, determining a target model from a preset physical model, a pre-trained standpipe pressure prediction neural network, and a pre-trained drilling rig status prediction neural network according to the real-time status data;

[0054] Step S103, if the target model is the physical model, based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is determined by the physical model;

[0055] Step S104, if the target model is the standpipe pressure prediction neural network, inputting the real-time status data into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted;

[0056] Step S105, if the target model is the drilling rig state prediction neural network, inputting the real-time state data into the drilling rig state prediction neural network to obtain a predicted kinetic energy coefficient;

[0057] Step S106, using the predicted kinetic energy coefficient to replace the corresponding empirical kinetic energy coefficient in the physical model to obtain a hybrid physical model, and passing the hybrid physical model;

[0058] Step S107, based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model.

[0059] A method for predicting riser pressure provided in an embodiment of the present invention includes: acquiring real-time status data of a wellhead to be predicted; determining a target model from a preset physical model, a pre-trained riser pressure prediction neural network and a pre-trained rig status prediction neural network according to the real-time status data; if the target model is the physical model, determining the riser pressure prediction data of the wellhead to be predicted through the physical model based on the real-time status data; if the target model is the riser pressure prediction neural network, inputting the real-time status data into the riser pressure prediction neural network to obtain the riser pressure prediction data of the wellhead to be predicted; if the target model is the rig status prediction neural network, inputting the real-time status data into the riser pressure prediction neural network to obtain the predicted kinetic energy coefficient; replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model, and through the hybrid physical model; based on the real-time status data, through the hybrid physical model, obtaining the riser pressure prediction data of the wellhead to be predicted. By establishing a riser pressure prediction method based on a physical model and a riser pressure prediction method based on machine learning modeling, and comprehensively utilizing the advantages of the two methods to establish a hybrid model, the riser pressure prediction under a variety of different situations can be achieved, thereby predicting the occurrence of abnormal situations in advance.

[0060] For example 2, please refer to Figure 2 , Figure 2 The flowchart of Embodiment 2 of a method for predicting standpipe pressure of the present invention includes:

[0061] Step S201, obtaining real-time status data of the wellhead to be predicted;

[0062] Step S202, determining a target model from a preset physical model, a pre-trained standpipe pressure prediction neural network, and a pre-trained drilling rig status prediction neural network according to the real-time status data;

[0063] In an embodiment of the present invention, based on real-time status data, model selection is performed according to the following rules so as to select a suitable target model when performing riser pressure prediction: (1) When the well type is an exploratory well and there is no seismic and drilling data, a physical model is selected as the target model; (2) When the well type is a development well and there are more than 15 wells drilled in the block, a riser pressure prediction neural network model is selected as the target model; (3) When the well type is a development well and there are less than 10 wells drilled in the block, and when the pump is shut down for a long time or the drill string slips, a drilling rig status prediction neural network model is selected as the target model; (4) When there are too many false data due to outliers or sensor errors in the real-time status data, a physical model is selected as the target model.

[0064] It should be noted that the riser pressure is measured at the downstream high-pressure end of the pump. The drilling fluid circulation flow scenarios during the drilling process can be divided into three types: flow in the pipe (surface pipeline and inside the drill string), flow in the annulus and flow in the drill bit.

[0065] In the embodiment of the present invention, the physical model is specifically:

[0066] P p =ΔP s +ΔP d +ΔP b +ΔP a ;

[0067] Among them, P p is the standpipe pressure, ΔP s is the pressure loss in the surface pipeline. This prediction method ignores the small friction loss of the pump pressure on the surface pipeline, that is, ΔP s =0,ΔP d is the pressure loss in the drill string, ΔP b Pressure loss at the drill bit, ΔP a is the pressure loss in the annulus.

[0068] Specifically, the pressure loss in the drill string ΔP d and the pressure loss in the annulus ΔP a Size:

[0069]

[0070] Where Lm is the length of the drill pipe, τ ep is the viscosity friction coefficient, v p is the average flow velocity in the drill pipe. If we want to find ΔPa , D is the annulus diameter = annulus outer diameter - annulus inner diameter. For different annulus diameters, they should be calculated separately for each section and then added together. If ΔP is to be obtained d , D is the inner diameter of the drill pipe. If the drill pipe is different, it should be calculated separately for each section and then added. Viscosity friction coefficient τ ep The size is:

[0071]

[0072] Among them, τ y0 is the yield stress, τ y1 is the thixotropy related parameter (similar to Cheng's model), a and b are the kinetic energy coefficients, γ is the shear rate, and are the ratios of the drill string volume and annulus volume to the overall volume, t is the time from the start of flow, and ta is the delay time after the pressure wave propagates to the annulus. K is the drilling fluid viscosity coefficient, and m is the fluidity index. As time goes by, the shear stress will decay exponentially and eventually reach a steady state. When using this model to calculate pressure loss, it is necessary to estimate the kinetic energy coefficients a and b. Generally, the empirical value of a is 0.6, and the empirical value of b is 0.3. The kinetic energy coefficient can also be estimated by fitting or machine learning training. During the drilling process, when the drilling fluid parameters change, the curve fitting needs to be repeated.

[0073] Pressure loss at the drill bit ΔP b Size:

[0074]

[0075] Where Q is the volume flow rate, D n1 D is the diameter of the first water hole of the drill bit. n2 D is the diameter of the second water hole of the drill bit. nn is the diameter of the nth water hole of the drill bit.

[0076] In an optional embodiment, the training step of the standpipe pressure prediction neural network includes:

[0077] Acquire a riser pressure prediction training sample; the riser pressure training sample includes: historical state data and historical riser pressure at corresponding time; the historical state data includes: historical seismic data, historical drilling data and historical logging data;

[0078] The initially established prediction neural network is trained according to all the historical seismic data, the historical drilling data, the historical logging data and the historical standpipe pressure to obtain the standpipe pressure prediction neural network.

[0079] In the embodiment of the present invention, the historical state data of the block where the wellhead to be predicted is obtained from the database of different blocks constructed by collecting drilling data, logging data (flow rate, rotation speed, etc.), logging data and seismic data (layer velocity, etc.) of the blocks where different wellheads are located, that is, historical drilling data, historical logging data, historical logging data, historical seismic data and the historical standpipe pressure of the corresponding time as training samples, and the data is preprocessed and converted using the preprocessing library of Scikit-Learn, and then the training samples are divided into sequential intervals, from the standpipe pressure fluctuation to the steady-state response, and these sequence intervals are modeled as a unique hot encoding and included in the training of the network. Then, an initialized standpipe pressure prediction neural network is established, and the initialized standpipe pressure prediction neural network is trained according to the training samples to obtain the standpipe pressure prediction neural network.

[0080] In the specific implementation, the network model used for prediction during network training uses a feedforward neural network and RBF neural network algorithm, which is divided into input layer, hidden layer and output layer. The input layer node data includes parameters such as pump flow, pump pressure peak, rotation speed, layer speed, drill bit depth and well depth to train the network and predict the riser pressure. For the training samples, the data related to the input layer nodes of a certain block are selected, and various neural network configurations are analyzed. The data is divided into a training and testing ratio of 4:1 to train a single hidden layer neural network.

[0081] In an optional embodiment, the training step of the drilling rig prediction neural network includes:

[0082] Determine the first riser pressure prediction result corresponding to all the historical state data through a preset physical model;

[0083] Determining the second riser pressure prediction results corresponding to all the historical state data through a preset riser pressure prediction neural network;

[0084] The initially established drilling rig state detection model is trained according to the historical drilling rig state data, the first riser pressure prediction result and the second riser pressure prediction result to obtain a drilling rig state prediction neural network.

[0085] Step S203, if the target model is the physical model, based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is determined by the physical model;

[0086] Step S204: if the target model is the standpipe pressure prediction neural network, input the real-time status data into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted;

[0087] Step S205, if the target model is the drilling rig state prediction neural network, inputting the real-time state data into the drilling rig state prediction neural network to obtain a predicted kinetic energy coefficient;

[0088] Step S206, using the predicted kinetic energy coefficient to replace the corresponding empirical kinetic energy coefficient in the physical model to obtain a hybrid physical model, and passing the hybrid physical model;

[0089] Step S207, acquiring real-time status data of multiple adjacent wells, and determining the riser pressure prediction data of all adjacent wells through the hybrid physical model;

[0090] Step S208, using all the prediction data of the riser pressures of the adjacent wells to perform curve fitting on the hybrid physical model, to obtain the prediction data of the riser pressure of the wellhead to be predicted.

[0091] In an embodiment of the present invention, based on the physical model and the neural network for predicting the riser pressure, the kinetic energy coefficients a and b in the physical model are trained through the input layer data of the machine learning model in the block where the wellhead to be predicted is located and applied to the prediction of the riser pressure of the neighboring wells in the block. Then, the proposed model is curve-fitted using the riser pressure prediction data of the neighboring wells to estimate the kinetic energy coefficient, and then the riser pressure value of the current well is predicted. Then, the kinetic energy coefficient is estimated using the measured riser pressure of the drilled section of the current well to predict the riser pressure value of the next section of the current well, until the riser pressure prediction of the panoramic section is completed. In this way, the degree of change of the riser pressure of the entire well section can be predicted in real time, the riser pressure value can be predicted to the greatest extent, abnormal changes in the riser pressure can be discovered in a timely manner, risks can be discovered at the first time, and effective treatment measures can be taken to minimize losses.

[0092] A method for predicting riser pressure provided in an embodiment of the present invention includes: acquiring real-time status data of a wellhead to be predicted; determining a target model from a preset physical model, a pre-trained riser pressure prediction neural network and a pre-trained rig status prediction neural network according to the real-time status data; if the target model is the physical model, determining the riser pressure prediction data of the wellhead to be predicted through the physical model based on the real-time status data; if the target model is the riser pressure prediction neural network, inputting the real-time status data into the riser pressure prediction neural network to obtain the riser pressure prediction data of the wellhead to be predicted; if the target model is the rig status prediction neural network, inputting the real-time status data into the riser pressure prediction neural network to obtain the predicted kinetic energy coefficient; replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model, and through the hybrid physical model; based on the real-time status data, through the hybrid physical model, obtaining the riser pressure prediction data of the wellhead to be predicted. By establishing a riser pressure prediction method based on a physical model and a riser pressure prediction method based on machine learning modeling, and comprehensively utilizing the advantages of the two methods to establish a hybrid model, the riser pressure prediction in a variety of different situations can be achieved, and the occurrence of abnormal situations can be predicted in advance. The present invention can predict the degree of change of the riser pressure of the entire well section in real time during the drilling construction process, can predict the riser pressure value to the greatest extent, and promptly discover abnormal changes in the riser pressure, discover risks at the first time and take effective treatment measures to minimize losses.

[0093] Example 3, taking the drilling data in the northwest region as an example:

[0094] Well 1 has been drilled to a depth of 5,000 meters. There are 18 neighboring wells with acceptable data quality. At this time, the standpipe pressure prediction neural network based on machine learning is automatically selected as the target model, and the standpipe pressure values ​​of 5,000 to 7,000 meters are obtained without any abnormalities. Drilling can continue according to the existing design, providing effective protection for the site.

[0095] Example 4, also taking the drilling data in the northwest region as an example:

[0096] Well 2 has not been explored, and there is no seismic data, logging data, etc. There is a small amount of drilling data. The designed well depth is 7,000 meters. The physical model is automatically selected as the target model, and the kinetic energy coefficients a and b are empirical values. The riser pressure value from 0 to 7,000 meters is obtained. It is found that the predicted value of the riser pressure at 6,000 meters fluctuates greatly. According to the analysis, the riser pressure value is calculated again after the drilling fluid performance of this section is readjusted, and this section returns to normal. This method can timely and accurately predict the abnormal well section, which plays a guiding role in the design.

[0097] Embodiment 5 is a system for predicting standpipe pressure of the present invention, comprising:

[0098] The data import and automatic acquisition module is used to import seismic data, drilling data and logging data (automatic acquisition), and automatically store these data into the database using the storage function of the time series database;

[0099] A riser pressure prediction module based on a physical model is used to input calculation parameters, call a riser pressure prediction method based on a physical model, and display the output results in a visual manner and store them in a database;

[0100] The riser pressure prediction module based on machine learning modeling is used to train the model through machine learning according to the data in the database, and predict the riser pressure according to the measured data;

[0101] The model automatic selection module is used to automatically select the corresponding model for riser pressure prediction according to the rules and store the relevant data of the calculation results into the database.

[0102] Example 6, please refer to Figure 3 , Figure 3 The structure diagram of a standpipe pressure prediction device embodiment of the present invention includes:

[0103] The acquisition module 501 is used to acquire the real-time status data of the wellhead to be predicted;

[0104] A target model determination module 502 is used to determine a target model from a preset physical model, a pre-trained standpipe pressure prediction neural network and a pre-trained drilling rig state prediction neural network according to the real-time status data;

[0105] A first prediction module 503 is used to determine the riser pressure prediction data of the wellhead to be predicted based on the real-time status data and through the physical model if the target model is the physical model;

[0106] A second prediction module 504 is configured to input the real-time status data into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted if the target model is the standpipe pressure prediction neural network;

[0107] A third prediction module 505 is used for inputting the real-time status data into the drilling rig status prediction neural network to obtain a predicted kinetic energy coefficient if the target model is the drilling rig status prediction neural network; and

[0108] Replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model; and,

[0109] Based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model.

[0110] In an optional embodiment, it also includes:

[0111] A sample data acquisition module is used to acquire a riser pressure prediction training sample; the riser pressure training sample includes: historical state data and historical riser pressure at corresponding time; the historical state data includes: historical seismic data, historical drilling data and historical logging data;

[0112] The target neural network training module is used to train the initially established prediction neural network based on all the historical seismic data, the historical drilling data, the historical logging data and the historical standpipe pressure to obtain the standpipe pressure prediction neural network.

[0113] In an optional embodiment, it also includes:

[0114] A first prediction result determination module, used to determine the first riser pressure prediction result corresponding to all the historical state data through a preset physical model;

[0115] A second prediction result determination module, used to determine the second riser pressure prediction results corresponding to all the historical state data through a preset riser pressure prediction neural network;

[0116] The drilling rig state prediction neural network generation module is used to train the initially established drilling rig state detection model according to the historical drilling rig state data, the first riser pressure prediction result and the second riser pressure prediction result to obtain the drilling rig state prediction neural network.

[0117] In an optional embodiment, the third prediction module 505 includes:

[0118] A submodule for determining the prediction data of the riser pressure of adjacent wells, used for acquiring the real-time status data of a plurality of adjacent wells, and determining the prediction data of the riser pressure of all the adjacent wells through the hybrid physical model;

[0119] The curve fitting submodule is used to perform curve fitting on the hybrid physical model using all the prediction data of the riser pressures of the adjacent wells to obtain the prediction data of the riser pressure of the wellhead to be predicted.

[0120] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of a method for predicting standpipe pressure as described in any of the above embodiments.

[0121] An embodiment of the present invention further provides a computer storage medium having a computer program stored thereon. When the computer program is executed by the processor, the steps of a method for predicting standpipe pressure as described in any of the above embodiments are implemented.

[0122] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0123] In the several embodiments provided in the present application, it should be understood that the methods, devices, electronic devices and storage media disclosed in the present invention can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0124] 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, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0125] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0126] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a readable storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned readable storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0127] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting standpipe pressure, characterized in that: include: Obtain real-time status data of the wellhead to be predicted; Determining a target model from a pre-set physical model, a pre-trained riser pressure prediction neural network, and a pre-trained rig status prediction neural network according to the real-time status data; If the target model is the physical model, based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is determined by the physical model; If the target model is the standpipe pressure prediction neural network, the real-time status data is input into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted; If the target model is the drilling rig state prediction neural network, the real-time state data is input into the drilling rig state prediction neural network to obtain a predicted kinetic energy coefficient; Replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model, and passing the hybrid physical model; Based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model.

2. The method for predicting standpipe pressure according to claim 1, characterized in that: The training steps of the standpipe pressure prediction neural network include: Acquire a riser pressure prediction training sample; the riser pressure training sample includes: historical state data and historical riser pressure at corresponding time; the historical state data includes: historical seismic data, historical drilling data and historical logging data; The initially established prediction neural network is trained according to all the historical seismic data, the historical drilling data, the historical logging data and the historical standpipe pressure to obtain the standpipe pressure prediction neural network.

3. The method for predicting standpipe pressure according to claim 2, characterized in that: The training steps of the drilling rig prediction neural network include: Determine the first riser pressure prediction result corresponding to all the historical state data through a preset physical model; Determining the second riser pressure prediction results corresponding to all the historical state data through a preset riser pressure prediction neural network; The initially established drilling rig state detection model is trained according to the historical drilling rig state data, the first riser pressure prediction result and the second riser pressure prediction result to obtain a drilling rig state prediction neural network.

4. The method for predicting standpipe pressure according to claim 2, characterized in that: Based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model, including: Acquire real-time status data of multiple adjacent wells, and determine the riser pressure prediction data of all adjacent wells through the hybrid physical model; The hybrid physical model is curve-fitted using all the predicted data of the riser pressures of the adjacent wells to obtain the predicted data of the riser pressure of the wellhead to be predicted.

5. A standpipe pressure prediction device, characterized in that: include: An acquisition module is used to acquire real-time status data of the wellhead to be predicted; A target model determination module, used to determine the target model from a preset physical model, a pre-trained standpipe pressure prediction neural network and a pre-trained drilling rig state prediction neural network according to the real-time status data; A first prediction module, configured to determine the riser pressure prediction data of the wellhead to be predicted based on the real-time status data and through the physical model if the target model is the physical model; A second prediction module is used for inputting the real-time status data into the standpipe pressure prediction neural network to obtain the standpipe pressure prediction data of the wellhead to be predicted if the target model is the standpipe pressure prediction neural network; a third prediction module, for inputting the real-time status data into the drilling rig status prediction neural network to obtain a predicted kinetic energy coefficient if the target model is the drilling rig status prediction neural network; and Replacing the corresponding empirical kinetic energy coefficient in the physical model with the predicted kinetic energy coefficient to obtain a hybrid physical model; and, Based on the real-time status data, the riser pressure prediction data of the wellhead to be predicted is obtained through the hybrid physical model.

6. The standpipe pressure prediction device according to claim 5, characterized in that: Also includes: A sample data acquisition module is used to obtain standpipe pressure prediction training samples; The standpipe pressure training sample includes: historical state data and historical standpipe pressure at corresponding time; the historical state data includes: historical seismic data, historical drilling data and historical logging data; The target neural network training module is used to train the initially established prediction neural network based on all the historical seismic data, the historical drilling data, the historical logging data and the historical standpipe pressure to obtain the standpipe pressure prediction neural network.

7. The standpipe pressure prediction device according to claim 6, characterized in that: Also includes: A first prediction result determination module, used to determine the first riser pressure prediction result corresponding to all the historical state data through a preset physical model; A second prediction result determination module, used to determine the second riser pressure prediction results corresponding to all the historical state data through a preset riser pressure prediction neural network; The drilling rig state prediction neural network generation module is used to train the initially established drilling rig state detection model according to the historical drilling rig state data, the first riser pressure prediction result and the second riser pressure prediction result to obtain the drilling rig state prediction neural network.

8. The standpipe pressure prediction device according to claim 6, characterized in that: The third prediction module comprises: A submodule for determining the prediction data of the riser pressure of adjacent wells, used for acquiring the real-time status data of a plurality of adjacent wells, and determining the prediction data of the riser pressure of all the adjacent wells through the hybrid physical model; The curve fitting submodule is used to perform curve fitting on the hybrid physical model using all the prediction data of the riser pressures of the adjacent wells to obtain the prediction data of the riser pressure of the wellhead to be predicted.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 4 is executed.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 4 is executed.

Images