A real-time optimization modeling method for drilling speed based on core images

Abstract

The present invention discloses a real-time optimization modeling method for drilling speed based on core images. The model is established with rock images and drilling parameters as inputs to achieve real-time optimization of drilling operability parameters and achieve the purpose of optimal drilling speed. First, the features of the core image are extracted using mathematical methods and convolutional neural network methods at the same time; then the image features and drilling parameters are input into an artificial neural network together to establish a drilling speed model; finally, an optimization algorithm is used to obtain the optimal drilling parameters such as drilling pressure, rotation speed, pump volume, mud, etc., and the on-site drilling parameters are updated with the optimal drilling parameters to achieve the optimal drilling speed. The method proposed in this article overcomes the lag of lithology testing, and can optimize drilling parameters in real time through core images obtained at the core drilling site to increase the drilling speed.

Description

Technical Field

[0001] The invention relates to the field of geological exploration, and in particular to a real-time optimization modeling method for drilling speed based on core images. Background Art

[0002] Core drilling is an important means of geological exploration. The drilling rate is the penetration of rock per unit of pure drilling time and is one of the important indicators of drilling engineering. Establishing a drilling rate model and increasing the drilling rate can reduce the cost of core drilling. The rock performance parameters in the current drilling rate model are all obtained through laboratory testing, which causes the establishment of the drilling rate model to lag behind the on-site drilling project and make it impossible to optimize the drilling parameters at the drilling site in real time. Summary of the invention

[0003] In view of this, the present invention provides a real-time optimization modeling method for drilling speed based on core images, comprising the following steps:

[0004] Step 1: Count the pixel values ​​of the core image to obtain a histogram, set a threshold, set pixels greater than the threshold to 1, and pixels less than the threshold to 0. Record the continuous area with a pixel value of 0 as a core block, count the number of core blocks n, and calculate the number of pixels s in each core block. n and the core crushing characteristics T;

[0005] Step 2: Count the pixel values ​​of the core image, obtain the three pixel values ​​with the largest proportion of the core image pixels, and obtain the peak pixel characteristics of the core image;

[0006] Step 3: Detect straight lines on the core image, merge two parallel straight lines whose distance difference is less than △ρ, count the number, length and angle of the straight lines, and obtain the texture characteristics of the core image;

[0007] Step 4: Normalize the core image and input it into the convolutional neural network to extract the hidden features of the core;

[0008] Step 5: normalize the drilling parameters and input them into the first artificial neural network for pre-training, wherein the number of hidden layers of the first artificial neural network is in the range of a1 to a2, and the number of neurons in the hidden layer is in the range of b1 to b2, wherein a1, a2, b1, and b2 are preset values, and the weights of the first artificial neural network are saved;

[0009] Step 6: The core crushing features, peak pixel features, core image texture features, core hidden features extracted in steps 1-4 and the drilling parameters in step 5 are input into a second artificial neural network, the number of hidden layers of the second artificial neural network is in the range of a3 to a4, the number of hidden layer neurons is in the range of b3 to b4, wherein a3, a4, b3, and b4 are preset values, the weights of the first artificial neural network pre-trained with the drilling parameters in step 5 are loaded to form a joint model, the joint model is trained, and a drilling speed model is obtained;

[0010] Step 7: Input the core image acquired in real time into the joint model described in step 6, and use the maximum drilling speed as the objective function to obtain the optimal drilling operability parameters through the particle swarm optimization algorithm.

[0011] Furthermore, in step 1, the calculation formula of the crushing characteristic T is:

[0012]

[0013] Where L is the core length, S is the total number of pixels in the core block, S = s1 + s2 + ... + s n , the closer the core crushing characteristics are to 0, the more crushed the core is.

[0014] Furthermore, in step 3, a coordinate system is established with the point at the lower left corner of the image as the origin, the horizontal direction is the abscissa, and the vertical direction is the ordinate, and the Hough transform equation ρ=x cosθ+y sinθ is used to detect the straight line of the core image, wherein ρ is the distance from the straight line to the origin, θ is the angle between the straight line and the vertical direction, x represents the abscissa value of the foot of the perpendicular from the origin to the straight line, and y represents the ordinate value of the foot of the perpendicular from the origin to the straight line.

[0015] Furthermore, in step 4, the core image is normalized, specifically: the core image is reset and cropped to the same size, and then the tensor of the image input channel is transformed into a standard normal distribution, and the calculation formula is as follows:

[0016]

[0017] Where input[channel] is the input image channel tensor, mean[channel] is the mean of the selected standard normal distribution, and std[channel] is the variance of the selected standard normal distribution.

[0018] Furthermore, the drilling parameters in step 5 include drill bit depth; the drilling operability parameters include drilling pressure, rotation speed, pump volume, and mud performance; the drilling parameters are normalized, and the calculation formula is:

[0019]

[0020] Where: x i represents drilling parameters, x max Indicates the maximum value of drilling parameters, x min Indicates the minimum value of drilling parameters.

[0021] The beneficial effects brought by the technical solution provided by the present invention are:

[0022] Compared with the existing technology, it overcomes the disadvantage of lagging core performance testing. Through the acquired core images, the drilling speed is modeled in real time, and the drilling operability parameters are optimized in real time, achieving the effect of improving the drilling speed in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a real-time optimization modeling method for drilling speed based on core images of the present invention;

[0024] Figure 2 are the drilling parameters and core images of the borehole in the embodiment of the present invention;

[0025] Figure 3 is a scatter plot of predicted values ​​and true values ​​in an embodiment of the present invention;

[0026] Figure 4 1 is a comparison diagram of the original value and the optimized value of the drilling speed for each round in the embodiment of the present invention;

[0027] Figure 5 1 is a comparison diagram of the original value and the optimized value of the rotation speed of each round in the embodiment of the present invention;

[0028] Figure 6 It is a comparison chart of the original value and the optimized value of the drilling pressure in each round in the embodiment of the present invention. DETAILED DESCRIPTION

[0029] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0030] refer to Figure 1 , Figure 1 The present invention is a flow chart of a real-time optimization modeling method for drilling speed based on core images.

[0031] The specific technical solution of the present invention is illustrated by testing the data of a core drilling hole in Gansu:

[0032] refer to Figure 2 , Figure 2 are the drilling parameters and core images of the borehole in the embodiment of the present invention.

[0033] Step 1: Count the pixel values ​​of the core image to obtain a histogram, set a threshold, set pixels greater than the threshold to 1, and pixels less than the threshold to 0. Record the continuous area with a pixel value of 0 as a core block, count the number of core blocks n, and calculate the number of pixels s in each core block. n and the core crushing characteristics T;

[0034] The calculation formula of the crushing characteristic T is:

[0035]

[0036] Where L is the core length, S is the total number of pixels in the core block, S = s1 + s2 + ... + s n , the closer the core crushing characteristics are to 0, the more crushed the core is.

[0037] Step 2: Count the pixel values ​​of the core image, obtain the three pixel values ​​with the largest proportion of the core image pixels, and obtain the peak pixel feature;

[0038] Step 3: Detect straight lines on the core image. When detecting straight lines, for thicker straight lines, multiple straight lines will be detected, so these straight lines need to be merged. Merge two parallel straight lines with a distance difference less than △ρ, count the number, length, and angle of the straight lines, and obtain the texture characteristics of the core image;

[0039] A coordinate system is established with the point at the lower left corner of the image as the origin, with the horizontal direction as the horizontal coordinate x and the vertical direction as the vertical coordinate y. The Hough transform equation ρ=x cosθ+y sinθ is used to detect the straight line of the core image, where ρ is the distance from the straight line to the origin of the image, θ is the angle with the vertical coordinate, x represents the horizontal coordinate value, and y represents the vertical coordinate value. In this embodiment, △ρ takes a value of 5um.

[0040] Step 4: Normalize the core image and input it into the convolutional neural network to extract the hidden features of the core;

[0041] Normalization of the core image is as follows: the core image is reset and cropped to the same size, and then the tensor of the image input channel is transformed into a standard normal distribution. The calculation formula is as follows:

[0042]

[0043] Where input[channel] is the input image channel tensor, mean[channel] is the mean of the selected standard normal distribution, and std[channel] is the variance of the selected standard normal distribution.

[0044] In this embodiment, the image size of the core is reset to (32×256), and then the tensor of the image input channel is transformed into a standard normal distribution, with a mean value of [0.485, 0.456, 0.406] and a variance of [0.229, 0.224, 0.225].

[0045] The normalized core image is input into the convolutional neural network Resnet-50 to extract 1 to 1000 core features. In this embodiment, 10 core features are extracted.

[0046] Step 5: Normalize the drilling parameters and the calculation formula is as follows

[0047]

[0048] Where: x i represents drilling parameters, x max is the maximum value of the drilling parameter, x min It is the minimum value of drilling parameters.

[0049] The normalized drilling parameters are input into the first artificial neural network for pre-training, wherein the number of hidden layers of the first artificial neural network is in the range of 1 to 100, and the number of neurons in the hidden layer is in the range of 1 to 10000. In this embodiment, the number of hidden layers of the first artificial neural network is 1, and the number of neurons in the hidden layer is 32. The weights of the artificial neural network are saved, and the drilling parameters include drill bit depth, drilling operability parameters: drilling pressure, speed, pump volume, mud performance parameters, mechanical characteristic parameters: drill bit size, drill bit mechanical performance parameters, etc.

[0050] Step 6: The core crushing features, peak pixel features, core image texture features, core hidden features extracted in steps 1-4 and the normalized drilling parameters obtained in step 5 are input into the second artificial neural network. In this embodiment, the number of hidden layers of the second artificial neural network is 1, and the number of hidden layer neurons is 32. The weights of the first artificial neural network pre-trained with drilling parameters in step 5 are loaded to form a joint model, and the joint model is trained to obtain a drilling speed model.

[0051] Model results after training Figure 3 , Figure 3 is a scatter plot of the predicted value and the true value in the embodiment of the present invention, and the R of the actual drilling speed and the predicted drilling speed of the model on the training set 2 The R on the test set is 0.99. 2 It is 0.91, and the model has high accuracy.

[0052] Step 7: During the drilling process, the core images obtained in real time are acquired by photographic equipment, and the core images acquired in real time are input into the joint model described in step 6. The optimal drilling operability parameters (rotation speed and drilling pressure) are obtained by particle swarm optimization algorithm with maximizing the drilling speed as the objective function.

[0053] During the drilling process, the core images are acquired in real time using photographic equipment, and the core images are input into the joint model of the convolutional neural network and the second artificial neural network described in step 6. The drilling operability parameters (rotation speed and drilling pressure) are optimized by the particle swarm optimization algorithm with maximizing the drilling speed as the objective function.

[0054] Drilling speed reference before and after optimization Figure 4 , Figure 4 The comparison chart of the original value and optimized value of the drilling speed of each round in the embodiment of the present invention is shown in the figure; it can be seen from the figure that the average drilling speed is increased by 19%, and the drilling operability drilling parameters rotation speed and drilling pressure reference before and after optimization Figure 5 and Figure 6 , Figure 5 1 is a comparison diagram of the original value and the optimized value of the rotation speed of each round in the embodiment of the present invention; Figure 6 It is a comparison chart of the original value and the optimized value of the drilling pressure in each round in the embodiment of the present invention.

[0055] Compared with the current drilling speed model, this method can optimize the drilling operability parameters and maximize the drilling speed in real time through the core images obtained by drilling. The process of core performance testing is omitted, making the drilling speed optimization real-time.

[0056] Compared with the existing technology, the present invention overcomes the disadvantage of delayed core performance testing, and can achieve the effect of improving the drilling speed in real time by modeling the drilling speed in real time through the acquired core image and optimizing the drilling operability parameters in real time.

[0057] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time optimization modeling method for drilling speed based on core images, characterized in that: The following steps are involved: Step 1: Count the pixel values ​​of the core image to obtain a histogram, set a threshold, set pixels greater than the threshold to 1, and pixels less than the threshold to 0. Record the continuous area with a pixel value of 0 as a core block, count the number of core blocks n, and calculate the number of pixels s in each core block. n and the core crushing characteristics T; Step 2: Count the pixel values ​​of the core image, obtain the three pixel values ​​with the largest proportion of the core image pixels, and obtain the peak pixel characteristics of the core image; Step 3: Detect straight lines on the core image, merge two parallel straight lines whose distance difference is less than △ρ, count the number, length and angle of the straight lines, and obtain the texture characteristics of the core image; Step 4: Normalize the core image and input it into the convolutional neural network to extract the core hidden features; Step 5: normalize the drilling parameters and input them into the first artificial neural network for pre-training, wherein the number of hidden layers of the first artificial neural network is in the range of a1 to a2, and the number of neurons in the hidden layer is in the range of b1 to b2, wherein a1, a2, b1, and b2 are preset values, and the weights of the first artificial neural network are saved; Step 6: The core crushing features, peak pixel features, core image texture features, core hidden features extracted in steps 1-4 and the drilling parameters in step 5 are input into a second artificial neural network, the number of hidden layers of the second artificial neural network is in the range of a3 to a4, the number of hidden layer neurons is in the range of b3 to b4, wherein a3, a4, b3, and b4 are preset values, the weights of the first artificial neural network pre-trained with the drilling parameters in step 5 are loaded to form a joint model, the joint model is trained, and a drilling speed model is obtained; Step 7: Input the core image acquired in real time into the joint model described in step 6, and use the maximum drilling speed as the objective function to obtain the optimal drilling operability parameters through the particle swarm optimization algorithm.

2. The method for real-time optimization modeling of drilling speed based on core images according to claim 1, characterized in that: In step 1, the calculation formula of the crushing characteristic T is: Where L is the core length, S is the total number of pixels in the core block, S = s1 + s2 + ... + s n , the closer the core crushing characteristics are to 0, the more crushed the core is.

3. The method for real-time optimization modeling of drilling speed based on core images according to claim 1, characterized in that: In step 3, a coordinate system is established with the point at the lower left corner of the image as the origin, the horizontal direction is the abscissa, and the vertical direction is the ordinate, and the Hough transform equation ρ=xcosθ+ysinθ is used to detect the straight line of the core image, where ρ is the distance from the straight line to the origin, θ is the angle between the straight line and the vertical direction, x represents the abscissa value of the foot of the perpendicular from the origin to the straight line, and y represents the ordinate value of the foot of the perpendicular from the origin to the straight line.

4. The method for real-time optimization modeling of drilling speed based on core images according to claim 1, characterized in that: In step 4, the core image is normalized, specifically: the core image is reset and cropped to the same size, and then the tensor of the image input channel is transformed into a standard normal distribution. The calculation formula is as follows: Where input[channel] is the input image channel tensor, mean[channel] is the mean of the selected standard normal distribution, and std[channel] is the variance of the selected standard normal distribution.

5. The method for real-time optimization modeling of drilling speed based on core images according to claim 1, characterized in that: The drilling parameters in step 5 include the drill bit depth; the drilling operability parameters include drilling pressure, rotation speed, pump volume, and mud performance; the drilling parameters are normalized, and the calculation formula is: Where: x i represents drilling parameters, x max Indicates the maximum value of drilling parameters, x min Indicates the minimum value of drilling parameters.

Images