Drilling parameter online optimization method

Abstract

The invention discloses a drilling parameter online optimization method, which is characterized in that a drilling parameter dynamic optimization problem is abstracted into a dynamic optimization problem changing along with time, and the specific method comprises the following steps: S1, data acquisition and preprocessing; s2, training a mechanical drilling speed prediction model based on an adaptive random forest algorithm to update a target function; s3, drilling parameter constraint conditions are updated; s4, the decision variable space is updated by considering the change range of the drilling parameters; s5, solving a drilling parameter optimization problem by using a particle swarm algorithm under a drilling parameter constraint condition to obtain optimized drilling parameters, namely bit pressure, rotating speed and displacement; and S6, the steps S1 to S5 are repeated, in the time space, the target function, the drilling parameter constraint condition and the decision variable space are updated, and the optimal drilling parameter scheme under the current drilling condition is solved. According to the drilling parameter online optimization method, the drilling parameters can be dynamically optimized according to drilling condition changes, and the method has important significance on intelligent drilling.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas drilling, in particular to an online optimization method for drilling parameters. Background Art

[0002] During the actual drilling process, due to the uncertainty and time-varying nature of drilling conditions, the mechanical drilling rate prediction model presents the characteristics of dynamic change. Constraints such as stick-slip vibration also show dynamic changes. The objective function and constraints of the mechanical drilling rate optimization problem change with the passage of time. There is a dynamic optimal solution under different drilling conditions. The solution to the problem will expand or shrink with time, and the new solution may become better or worse during the evolution of the problem.

[0003] The characteristic of the dynamic optimization problem of drilling parameters is that the drilling data arrives gradually, and the algorithm makes real-time decisions based on the currently known information at each time step. However, the static optimization design of drilling parameters can obtain all the information at the beginning of the solution, especially after the drilling conditions change. The real-time optimization of drilling parameters can only observe part of the drilling data at each time step, and cannot know all the information in advance.

[0004] To this end, the present invention integrates drilling engineering with artificial intelligence algorithms and proposes an online optimization method for drilling parameters, which can dynamically optimize drilling parameters according to changes in drilling conditions, which is of great significance for realizing intelligent drilling. Summary of the invention

[0005] The drilling parameter online optimization method provided by the present invention has the following specific technical solutions:

[0006] The core idea of ​​the online optimization method of drilling parameters of the present invention is to abstract the dynamic optimization problem of drilling parameters into a dynamic optimization problem that changes with time. Its goal is to find x(t) in the entire time-varying process so that F(x, t) is minimized under given constraints. The formula is as follows:

[0007] minF(x,t)=(f1(x,t),...,f mt (x,t) T

[0008] stg i (x,t)≤0,i=1,2,...,p

[0009] h j (x,t)=0,j=1,2,...,q

[0010] x∈Ω x ,t∈Ω t

[0011] Where: fmt (x, t) represents the mtth objective function at time t, specifically representing the mechanical drilling speed prediction model.

[0012] x=(x1,x2,...x nt ) T Represents the decision optimization variable of nt latitude. The value range of x changes with time. x=(x1,x2,...x nt ) T Specifically represents drilling parameters including drilling pressure, rotation speed and displacement.

[0013] F(x,t) represents the objective function vector evaluating the solution x at time t.

[0014] g i (x, t) represents the inequality constraint, which changes with time; h j (x, t) represents the equality constraint, which changes with time; i and j represent multiple constraints, which include but are not limited to drilling pressure constraint, rotation speed constraint, mechanical drilling speed constraint and stick-slip vibration constraint.

[0015] Ω x Represents the feasible decision variable space, specifically the decision variable space for drilling parameter adjustment.

[0016] Ω t It represents time space, specifically the time range for drilling parameter optimization.

[0017] The drilling parameter online optimization method of the present invention comprises the following steps:

[0018] S1. Acquire quasi-dynamic data based on drilling logs, and pre-process the real-time logging data at the drilling site.

[0019] The fixed-frequency drilling pressure, rotation speed, displacement, and mechanical drilling speed are obtained based on the real-time logging data at the drilling site, and the real-time drilling data is dynamically preprocessed, including the elimination of data anomalies and data smoothing and noise reduction.

[0020] The quasi-dynamic data obtained from the drilling log includes drilling tool assembly data, wellbore structure data, wellbore trajectory data and drilling fluid density data.

[0021] S2. Update the objective function of the mechanical drilling speed prediction model based on the adaptive random forest algorithm training; the details are as follows:

[0022] Based on the adaptive random forest algorithm, with drilling pressure, rotation speed and displacement as input feature variables and mechanical drilling speed as output feature variable, the preprocessed real-time drilling data is learned online to train and update the mechanical drilling speed prediction model, that is, to update the objective function f(x, t) in real time, which is expressed as:

[0023] f(x,t m )→f(x,t n )

[0024] f(x,t m ) represents t m The objective function at the moment; f(x,t n ) represents t n The objective function at time.

[0025] S3. Update drilling parameter constraints:

[0026] Considering the quasi-dynamic data such as drilling tool assembly data, wellbore structure data, wellbore trajectory data, drilling fluid density data and changes in formation properties, the constraint condition g is updated. i (x,t) and h j (x, t), calculate the critical drilling pressure, critical mechanical drilling speed and stick-slip vibration constraints. It is expressed as follows:

[0027] g i (x,t m )≤0→g i (x,t n )≤0

[0028] h j (x,t m )=0→h j (x,t n )=0

[0029] In the formula, g i (x,t m ) represents t m Inequality constraints at time; g i (x,t n ) represents t n Inequality constraints at time; h j (x,t m ) represents t m Equality constraints at time h j (x,t n ) represents t n The equality constraints at time.

[0030] S4. Update the decision variable space:

[0031] Considering the control range of drilling parameters and possible errors in the calculation model, the decision variable space is fine-tuned and updated within the range of existing data; it is expressed as follows:

[0032] x∈Ω xm →x∈Ω xn

[0033] In the formula, Ω xm Represents t m The decision variable space at the moment; Ω xn Represents t n The decision variable space at time .

[0034] S5. Particle Swarm Optimization

[0035] According to the established mechanical drilling speed prediction model, the particle swarm algorithm is used to solve the drilling parameter optimization problem under the drilling parameter constraints, and the optimized drilling parameters, namely drilling pressure, speed and displacement, are obtained.

[0036] S6. Repeat steps S1 to S5, update the objective function, drilling parameter constraints, and decision variable space in time and space, and solve the optimal drilling parameter solution under the current drilling conditions.

[0037] Compared with the prior art, the present invention is beneficial in that:

[0038] (1) The present invention integrates drilling engineering with artificial intelligence algorithms and proposes an online optimization method for drilling parameters, which can dynamically optimize drilling parameters according to changes in drilling conditions, and is of great significance for realizing intelligent drilling.

[0039] (2) The present invention can be directly applied on the basis of existing conventional drilling equipment without the need for additional downhole measurement tools.

[0040] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The figure is a flow chart of the online optimization method of drilling parameters of the present invention.

[0042] Figure 2 It is a schematic diagram of the principle of the online optimization method of drilling parameters of the present invention.

[0043] Figure 3 It is the online optimization result of drilling parameters in the embodiment. DETAILED DESCRIPTION

[0044] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0045] like Figure 1 and 2 As shown, the drilling parameter online optimization method of the present invention has the following detailed steps:

[0046] S1. Data acquisition and preprocessing

[0047] (1) Obtain fixed-frequency drilling pressure, rotation speed, displacement, and mechanical drilling speed based on real-time logging data at the drilling site, and dynamically pre-process the real-time drilling data, including the removal of data anomalies and data smoothing and noise reduction; (2) Obtain quasi-dynamic data based on the drilling log, including drill tool assembly data, wellbore structure data, wellbore trajectory data, and drilling fluid density data.

[0048] S2. Objective function update

[0049] Based on the adaptive random forest algorithm, with drilling pressure, rotation speed, and displacement as input feature variables and mechanical drilling speed as output feature variables, online learning of preprocessed real-time drilling data is used to train and update the mechanical drilling speed prediction model and update the objective function in real time. It can be expressed as:

[0050] f(x,t m )→f(x,t n )

[0051] f(x,t m ) represents t m The objective function at the moment; f(x,t n ) represents t n The objective function at time.

[0052] S3. Constraint update

[0053] Considering the quasi-dynamic data such as drilling tool assembly data, wellbore structure data, wellbore trajectory data, drilling fluid density data and changes in formation properties, the constraints are updated, and the critical drilling pressure, critical mechanical drilling speed, stick-slip vibration constraints, etc. are calculated. It can be expressed as:

[0054] g i (x,t m )≤0→g i (x,t n )≤0

[0055] h j (x,t m )=0→h j (x,t n )=0

[0056] In the formula, g i (x,t m ) represents t m Inequality constraints at time; g i (x,t n ) represents t n Inequality constraints at time; h j (x,t m ) represents t mEquality constraints at time h j (x,t n ) represents t n The equality constraints at the time. i and j represent some specific constraints.

[0057] Take the stick-slip vibration constraint condition as an example. The stick-slip vibration constraint condition is related to the drilling tool assembly, wellbore trajectory, drilling fluid density, wellbore structure, etc. Assume t m The drilling tool assembly, wellbore trajectory, drilling fluid density, and wellbore structure at the time can be expressed as data m , t m The stick-slip vibration constraint condition at the moment is: g(x,data m )≤0; at t n The drilling tool assembly, wellbore trajectory, drilling fluid density, and wellbore structure data at the moment have changed. m becomes data n , then t n The stick-slip vibration constraint condition at the moment is g(x,data n )≤0.

[0058] From m Time to t n The update of the stick-slip vibration constraint condition at time t can be expressed as:

[0059] g(x,data m )≤0→g(x,data n )≤0

[0060] S4. Decision variable space update

[0061] Considering the control range of drilling parameters and possible errors in the calculation model, the control range of decision variables is generally considered to be fine-tuned within the range of existing data. It can be expressed as:

[0062] x∈Ω xm →x∈Ω xn

[0063] In the formula, Ω xm Represents t m The decision variable space at the moment; Ω xn Represents t n The decision variable space at time .

[0064] S5. Particle Swarm Algorithm Optimization

[0065] According to the established mechanical penetration rate prediction model, the particle swarm algorithm is used to solve the drilling parameter optimization problem under the drilling parameter constraints, and the optimized drilling parameters, namely, drilling pressure, rotation speed and displacement, are obtained.

[0066] S6. Repeat steps S1 to S5, update the objective function, constraint conditions, and decision variable space in time and space, and solve the optimal drilling parameter solution under the current drilling conditions.

[0067] Specific application cases

[0068] The drilling parameter online optimization method of the present invention was applied and tested in the 1740m-1750m section of the PLH2 well. The test results are as follows: Figure 3 As shown in the figure, it can be seen that in the well section of 1740m to 1743m, the mechanical drilling speed increases with the increase of drilling pressure; this method shows that increasing drilling pressure is conducive to increasing the mechanical drilling speed.

[0069] At this time, the mechanical drilling speed prediction model, that is, the objective function, can be expressed as:

[0070] ROP=a*WOB α +b*RPM β +c*FLOW γ , obviously, a>0;

[0071] Wherein, ROP is the mechanical drilling speed, m / h; WOB is the drilling pressure, kN; RPM is the rotation speed, r / min; FLOW is the displacement, L / s; a, b, c are the equation coefficients, α, β, γ are the drilling pressure index, rotation speed index, and flow index respectively.

[0072] However, in the subsequent 1743m-1750m well section, the mechanical drilling speed increased significantly as the drilling pressure decreased.

[0073] At this time, the mechanical drilling speed prediction model, that is, the objective function, can be expressed as:

[0074] ROP=a*WOB α +b*RPM β +c*FLOW γ , obviously, a<0;

[0075] This is obviously because the drilling speed equation changes due to changes in the formation or other downhole conditions. However, as can be seen from the figure, the drilling pressure recommended by the algorithm also increases first and then decreases. In other words, the algorithm updates the mechanical drilling speed model to adapt to changes in drilling conditions, and can optimize drilling parameters under dynamic drilling conditions.

[0076] The above description is only a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of the technical solution of the present invention.

Claims

1. A drilling parameter online optimization method, characterized in that: The dynamic optimization problem of drilling parameters is abstracted into a dynamic optimization problem that changes with time. Its goal is to find x(t) in the entire time-varying process so that F(x, t) is minimized under given constraints. The formula is as follows: minF(x,t)=(f1(x,t),...,f mt (x,t)) T s.t.g i (x,t)≤0,i=1,2,...,p h j (x,t)=0,j=1,2,...,q x∈Ω x ,t∈Ω t Where: f mt (x, t) represents the mtth objective function at time t, specifically representing the mechanical drilling rate prediction model; x=(x1,x2,...x nt ) T represents the decision optimization variable of nt latitude, the value range of x changes with time, x=(x1,x2,...x nt ) T Specifically represents drilling parameters including drilling pressure, rotation speed, and displacement; F(x,t) represents the objective function vector evaluating the solution x at time t; g i (x, t) represents the inequality constraint, which changes with time; h j (x, t) represents the equality constraint, which changes with time; i and j represent multiple constraints, including but not limited to drilling pressure constraint, rotation speed constraint, mechanical drilling speed constraint and stick-slip vibration constraint; Ω x represents the decision variable space, specifically the decision variable space for drilling parameter adjustment; Ω t It represents time space, specifically the time range for drilling parameter optimization.

2. The method for online optimization of drilling parameters according to claim 1, characterized in that: The following steps are involved: S1. Acquire quasi-dynamic data based on drilling logs, and pre-process the real-time logging data at the drilling site; S2. Train the mechanical drilling speed prediction model based on the adaptive random forest algorithm and update the objective function: Based on the adaptive random forest algorithm, the real-time logging data of the drilling site is used as the input feature variable and the mechanical drilling rate is used as the output feature variable. The real-time drilling data after preprocessing is learned online to train and update the mechanical drilling rate prediction model, that is, to update the objective function f(x, t) in real time, which is expressed as: f(x,t m )→f(x,t n ) f(x,t m ) represents t m The objective function at the moment; f(x,t n ) represents t n The objective function at the moment; S3. Update drilling parameter constraints: Update the constraint g by considering quasi-dynamic data and changes in formation properties i (x,t) and h j (x, t), calculate the critical drilling pressure, critical mechanical drilling speed and stick-slip vibration constraints; S4. Update the decision variable space: Considering the control range of drilling parameters and the errors in the calculation model, the decision variable space is adjusted and updated within the range of existing data; S5. Particle Swarm Algorithm Optimization According to the established mechanical drilling speed prediction model, the particle swarm algorithm is used to solve the drilling parameter optimization problem under the drilling parameter constraints, and the optimized drilling parameters, namely drilling pressure, speed and displacement, are obtained. S6. Repeat steps S1 to S5, update the objective function, drilling parameter constraints, and decision variable space in time and space, and solve the optimal drilling parameter solution under the current drilling conditions.

3. The method for online optimization of drilling parameters according to claim 2, characterized in that: In step S1, the drilling pressure, rotation speed, displacement, and mechanical drilling speed of a fixed frequency are obtained according to the real-time logging data at the drilling site, and the real-time drilling data is dynamically preprocessed, including the elimination of data abnormal points and data smoothing and noise reduction.

4. The method for online optimization of drilling parameters according to claim 2, characterized in that: In step S1, the quasi-dynamic data includes drilling tool assembly data, wellbore structure data, wellbore trajectory data and drilling fluid density data.

5. The method for online optimization of drilling parameters according to claim 3, characterized in that: In step S2, drilling pressure, rotation speed and displacement are used as input characteristic variables, and mechanical drilling speed is used as output characteristic variable.

6. The method for online optimization of drilling parameters according to claim 2, characterized in that: In step S3, the constraint conditions are updated as follows: g i (x,t m )≤0→g i (x,t n )≤0 h j (x,t m )=0→h j (x,t n )=0 In the formula, g i (x,t m ) represents t m Inequality constraints at time; g i (x,t n ) represents t n Inequality constraints at time; h j (x,t m ) represents t m Equality constraints at time h j (x,t n ) represents t n The equality constraints at time.

7. The method for online optimization of drilling parameters according to claim 2, characterized in that: In step S4, the decision variable space is updated, which is expressed as follows: x∈Ω xm →x∈Ω xn In the formula, Ω xm Represents t m The decision variable space at the moment; Ω xn Represents t n The decision variable space at time.

Images