Small drilling rig multi-model drilling speed prediction method and device based on formation lithology classification

Abstract

The application discloses a small drilling rig multi-model drilling speed prediction method and equipment based on stratum lithology classification, considers the important factor of stratum lithology reflecting the drilling environment of the small drilling rig in the role of the drilling speed prediction, introduces stratum lithology information, establishes drilling speed prediction models of multiple lithologies, and is divided into three stages, in the first stage, according to actual engineering application experience, data is screened and resampled, then, the data is processed through wavelet filtering and normalization; in the second stage, three typical lithologies are selected from drilling data, and real-time drilling data is used for lithology classification; in the third stage, the lithology classification result of the second stage is used, drilling data of classification is introduced into the drilling speed model of the corresponding lithology for prediction, a good foundation is laid for intelligent control research in the drilling process of the small drilling rig, the problem that a single drilling speed model has low adaptability to different stratum lithologies is solved, and drilling speed prediction precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geological drilling, and particularly relates to a small drilling rig multi-model drilling speed prediction method and device based on formation lithology classification. BACKGROUND

[0002] At present, the drilling construction of the small drilling rig mainly relies on manual experience. In order to further improve the quality and efficiency of geological survey research work and promote the development of geological work in the direction of precision, speed and comprehensiveness, the development of high-end intelligent drilling technology plays an important role in the geological exploration and development of the small drilling rig. Drilling speed is a key indicator to measure drilling efficiency, and accurate drilling speed prediction is of great significance to the drilling process. However, the drilling process of the small drilling rig is characterized by complex geomechanical environment, nonlinearity, strong coupling and strong interference, which leads to low accuracy of the drilling speed model. Therefore, establishing a high-precision drilling speed model is an important basis for optimizing the efficiency of the drilling process of the small drilling rig. SUMMARY

[0003] In order to solve the above problems, the present application provides a small drilling rig multi-model drilling speed prediction method and device based on formation lithology classification, a small drilling rig multi-model drilling speed prediction method based on formation lithology classification, mainly comprising:

[0004] S1: According to the actual engineering application experience, the depth, drilling pressure, rotating speed, torque, standpipe pressure and drilling speed data are screened and resampled, and then the data is processed through wavelet filtering and normalization;

[0005] S2: The drilling pressure, rotating speed, torque, standpipe pressure and drilling speed data corresponding to the formation lithology are divided into a training set, a validation set and a test set after being processed by step S1; a lithology classification model is established based on the extreme learning machine and the four-fold cross-validation method, and the training set and the validation set data are used to train the lithology classification model, the input is the drilling pressure, the rotating speed, the torque, the standpipe pressure and the drilling speed, and the output is the lithology;

[0006] S3: The depth, drilling pressure and rotating speed data corresponding to the formation lithology are divided into a training set, a validation set and a test set, and a drilling speed prediction model for each lithology is established based on the extreme learning machine and the ten-fold cross-validation method, and the training set and the validation set data are used to train the drilling speed prediction model, the input is the depth, the drilling pressure and the rotating speed, and the output is the predicted drilling speed;

[0007] S4: The trained lithology classification model and the corresponding drilling speed prediction model are integrated to form a small drilling rig drilling speed prediction model based on formation lithology classification;

[0008] S5: The test set data is used to simulate and verify the drilling speed prediction method, and the small drilling rig drilling speed prediction model reaching the preset verification accuracy is applied to the actual engineering to obtain the predicted drilling speed in the actual engineering.

[0009] Further, step S1 specifically includes the following process:

[0010] 1) According to the actual engineering experience, the data of drilling pressure, rotation speed and drilling speed are screened under the condition that the well depth is equal to the drilling position, and the data screening needs to meet the following conditions: (1) the drilling pressure is greater than 0 (t), (2) the rotation speed is greater than 10 (r / min), (3) the drilling speed is greater than 0 (m / h) and less than 6 (m / h); and the data is resampled according to the set interval;

[0011] 2) The drilling data is filtered by using the wavelet threshold denoising method according to the following formula:

[0012]

[0013] Wherein a is the scale factor corresponding to t time, b is the displacement change amount corresponding to t time, is the wavelet base function.

[0014] 3) The drilling pressure, rotation speed, torque, standpipe pressure and drilling speed are normalized according to the following formula, so that the orders of magnitude of each index are the same:

[0015]

[0016] Wherein x * is the normalized data set, x is the real data set, min(x) is the smallest data in the data set, and max(x) is the largest data in the data set.

[0017] Further, the set interval in step 1) is 0.01 m.

[0018] Further, step S2 specifically includes the following process:

[0019] 1) The predicted value of the formation lithology is obtained by using the extreme learning machine method according to the following formula:

[0020]

[0021] T=H·β

[0022] Wherein g(x) is the activation function, ω i is the input weight, b i is the bias, T is the output matrix, β is the output weight, H is the hidden layer output matrix, and i=1, 2, 3,..., n.

[0023] 2) The input weight ω and the bias b of the model to be established are determined by using the four-fold cross-validation method, and then the lithology classification model is established.

[0024] Further, step S3 specifically includes the following process:

[0025] 1) Also using the extreme learning machine method, drilling speed prediction is carried out;

[0026] 2) The ten-fold cross-validation method is used to determine the input weight ω and the bias b of each to-be-established model, and then the drilling speed prediction model of each lithology is established.

[0027] 3) The prediction result of the lithology classification model is used as the switching condition of the drilling speed prediction model, the integration of the lithology classification model and the corresponding drilling speed prediction model is completed, the same stage drilling data is used as the judgment condition according to the lithology classification result, the input to the drilling speed model corresponding to the lithology is specified to carry out drilling speed prediction, so that the data corresponding to the lithology can be input to the corresponding drilling speed model.

[0028] Further, in the simulation verification in step S5, the calculation formula of the verification index of the test set is as follows:

[0029]

[0030]

[0031] Wherein, RMSE is the root mean square error, NRMSE is the normalized root mean square error, y i is the measured data, is the predicted data, and n is the sample number.

[0032] A small drilling rig multi-model drilling speed prediction device based on formation lithology classification, comprising: a processor and a storage device; the processor loads and executes the instructions and data stored in the storage device to realize a small drilling rig multi-model drilling speed prediction method based on formation lithology classification.

[0033] The technical scheme provided by the application has the beneficial effects that:

[0034] (1) The small drilling rig multi-model drilling speed prediction method based on formation lithology classification firstly performs screening, resampling, wavelet filtering and normalization and other preprocessing operations on drilling data, which can effectively improve the data quality and lay a good foundation for the subsequent lithology classification and drilling speed prediction modeling work;

[0035] (2) The small drilling rig multi-model drilling speed prediction method based on formation lithology classification adopts the extreme learning machine algorithm, uses the training set and the verification set data, and uses the four-fold cross-validation method to determine the hyperparameters of the model, and establishes the classification model of the formation lithology, which can use real-time drilling data to achieve the purpose of predicting the formation lithology;

[0036] (3) The small drilling rig multi-model drilling speed prediction method based on formation lithology classification provided by the present application uses the training set and verification set data corresponding to each lithology, adopts the extreme learning machine algorithm, determines the hyperparameters of the drilling speed model corresponding to each lithology by using the ten-fold cross-validation method, and integrates the small drilling rig drilling speed prediction model based on formation lithology classification. Then, the test set data is used to simulate and verify the prediction effect of the model, which is beneficial to the application of the present application in actual production. BRIEF DESCRIPTION OF DRAWINGS

[0037] The present application will be further described below in combination with the drawings and examples, wherein:

[0038] Figure 1 is a flowchart of the small drilling rig multi-model drilling speed prediction method based on formation lithology classification in the embodiment of the present application.

[0039] Figure 2 is a schematic diagram of drilling data denoising preprocessing in the embodiment of the present application.

[0040] Figure 3 is a schematic diagram of N-fold cross-validation in the embodiment of the present application.

[0041] Figure 4 is a small drilling rig lithology classification result graph of the test set in the embodiment of the present application.

[0042] Figure 5 is a small drilling rig drilling speed prediction result comparison graph in the embodiment of the present application.

[0043] Figure 6 is a hardware device working schematic diagram of the embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to have a clearer understanding of the technical features, objects and effects of the present application, the specific embodiments of the present application will be described in detail with reference to the drawings.

[0045] The embodiment of the present application provides a small drilling rig multi-model drilling speed prediction method and device based on formation lithology classification, and a small drilling rig for drilling within 500m.

[0046] Please refer to Figure 1 , Figure 1is a flow chart of a small rig multi-model drilling speed prediction method based on formation lithology classification in the embodiments of the present application, which divides the modeling process into three stages. In the first stage, the data is filtered and resampled according to the actual engineering application experience. Then, the data is processed through wavelet filtering and normalization. In the second stage, based on three typical lithologies including lithology A (limestone), lithology B (carbonaceous dolomite) and lithology C (carbonaceous limestone), the corresponding drilling data is selected, the depth, drilling pressure, rotation speed, torque, standpipe pressure and drilling speed data are divided into training set and test set, four-fold cross-validation is adopted, the training set is divided into 4 parts, 3 parts are new training set and 1 part is validation set, the extreme learning machine algorithm is used, the depth, drilling pressure, rotation speed, torque, standpipe pressure and drilling speed are used as model input, and the prediction results of the three lithologies are used as model output to establish the lithology classification model. In the third stage, the training set data is used, the ten-fold cross-validation method is adopted, the training set data of each lithology is divided into 10 equal parts, 9 parts are new training set and 1 part is validation set, the extreme learning machine algorithm is used, the depth, drilling pressure and rotation speed are used as model input, and the drilling speed prediction value is used as model output to establish the drilling speed prediction model corresponding to each lithology, and the lithology classification model and the corresponding drilling speed prediction model are integrated to form the drilling speed prediction model based on real-time drilling process parameter formation lithology classification. Finally, the test set data is used for simulation verification. The specific steps are as follows:

[0047] (1) The drilling data is preprocessed, and the comparison results before and after data denoising are shown in the figures in Figure 2 , the training set and validation set data based on lithology classification are shown in Table 1:

[0048] Table 1 Training set and validation set lithology samples

[0049] Name Number Lithology A 14 Lithology B 22 Lithology C 5

[0050] (2) The training and verification process of lithology classification and drilling speed prediction is shown in Figure 3 , the four-fold cross-validation method and the ten-fold cross-validation method are used to determine the model hyperparameters in the training process of lithology classification and drilling speed prediction, respectively.

[0051] (3) The test set data is used to test the small rig lithology classification model, as shown in Figure 4 , the final lithology classification accuracy can reach 74.21%.

[0052] (4) The test set is used to test the small rig drilling speed prediction, and the test results are shown in Table 2 and Figure 5The three comparison methods are the method provided in the application, the limit learning machine drilling speed prediction without lithology classification and the support vector regression drilling speed prediction using the limit learning machine for lithology classification.

[0053] Table 2 Comparison of small rig drilling speed prediction accuracy

[0054] Method RMSE NRMSE Proposed method 67.59％ 21.21％ Extreme learning machine drilling speed prediction without lithology classification 74.63％ 23.42％ Extreme learning machine lithology classification + support vector regression drilling speed prediction 72.33％ 22.70％

[0055] The application establishes a drilling speed prediction model for each lithology by fully considering that different stratum lithologies have different influences on the same drilling speed prediction model, and classifies the lithology by using real-time drilling data, so that the real-time acquisition of lithology information is ensured, and the drilling speed model has high modeling accuracy and strong generalization ability.

[0056] Please refer to Figure 6 , Figure 6 is a hardware device working schematic diagram of the embodiment of the application, and specifically comprises: a small rig multi-model drilling speed prediction device 401 based on stratum lithology classification, a processor 402 and a storage device 403.

[0057] The small rig multi-model drilling speed prediction device 401 based on stratum lithology classification: the small rig multi-model drilling speed prediction device 401 based on stratum lithology classification realizes the small rig multi-model drilling speed prediction method based on stratum lithology classification.

[0058] The processor 402: the processor 402 loads and executes instructions and data in the storage device 403 to realize the small rig multi-model drilling speed prediction method based on stratum lithology classification.

[0059] The storage device 403: the storage device 403 stores instructions and data; and the storage device 403 is used to realize the small rig multi-model drilling speed prediction method based on stratum lithology classification.

[0060] The application has the following beneficial effects:

[0061] (1) The small rig multi-model drilling speed prediction method based on stratum lithology classification firstly performs preprocessing operations such as screening, resampling, wavelet filtering and normalization on drilling data, so that the data quality is effectively improved, and a good foundation is laid for the following lithology classification and drilling speed prediction modeling work.

[0062] (2) A small drilling rig multi-model drilling speed prediction method based on formation lithology classification, which adopts an extreme learning machine algorithm, uses training set and verification set data, determines the hyperparameters of the model by using a four-fold cross-validation method, and establishes a formation lithology classification model, which can achieve the purpose of predicting formation lithology by using real-time drilling data;

[0063] (3) A small drilling rig multi-model drilling speed prediction method based on formation lithology classification, which uses training set and verification set data corresponding to each lithology, adopts an extreme learning machine algorithm, determines the hyperparameters of the drilling speed model corresponding to each lithology by using a ten-fold cross-validation method, and integrates them into a small drilling rig drilling speed prediction model based on formation lithology classification. Then, the test set data is used to simulate and verify the prediction effect of the model, which is conducive to the application of the present application in actual production.

[0064] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A multi-model drilling rate prediction method for small drilling rigs based on stratigraphic lithology classification, characterized in that: include: S1: Based on practical engineering application experience, the data on well depth, drilling pressure, rotation speed, torque, stand pressure and drilling speed are screened and resampled. Then, the data are processed by wavelet filtering and normalization. S2: After processing the data on drilling pressure, rotation speed, torque, stand pressure and drilling speed corresponding to the formation lithology in step S1, the data are divided into training set, validation set and test set; a lithology classification model is established based on extreme learning machine and four-fold cross-validation method, and the lithology classification model is trained using the training set and validation set data. Its input is drilling pressure, rotation speed, torque, stand pressure and drilling speed, and the output is lithology. S3: Divide the well depth, drilling pressure and rotation speed data corresponding to the formation lithology into training set, validation set and test set. Based on extreme learning machine and ten-fold cross-validation method, establish a drilling speed prediction model for each lithology. Use the training set and validation set data to train the drilling speed prediction model. Its input is well depth, drilling pressure and rotation speed, and the output is the predicted drilling speed. S4: Integrate the trained lithology classification model and the corresponding drilling speed prediction model to form a small drilling rig drilling speed prediction model based on formation lithology classification. S5: Use test set data to simulate and verify the drilling speed prediction method. Apply the small drilling rig drilling speed prediction model that has reached the preset verification accuracy to the actual project to obtain the actual predicted drilling speed.

2. The method for predicting drilling speed of small drilling rigs based on formation lithology classification as described in claim 1, characterized in that: Step S1 in detail The process includes the following: 1) Based on actual engineering experience, under the condition that the well depth is equal to the drilling position, data screening is performed on the drilling pressure, rotation speed and drilling speed. The data screening needs to meet the following conditions: (1) drilling pressure is greater than 0 (t), (2) rotation speed is greater than 10 (r / min), (3) drilling speed is greater than 0 (m / h) and less than 6 (m / h); and the data is resampled according to the set interval. 2) Filter the drilling data (drilling pressure, rotational speed, torque, stand pressure, and drilling speed) using the wavelet threshold denoising method according to the following formula: Where a is the scale factor at time t, and b is the displacement change at time t. These are wavelet basis functions; 3) Normalize the drilling pressure, rotational speed, torque, stand pressure, and drilling speed according to the following formula to make each index the same order of magnitude: Where x * The dataset is normalized, x is the actual dataset, min(x) is the smallest data point in the dataset, and max(x) is the largest data point in the dataset.

3. The method for predicting drilling speed of small drilling rigs based on formation lithology classification as described in claim 2, characterized in that: The interval set in step 1) is 0.01m.

4. The method for predicting drilling speed of small drilling rigs based on formation lithology classification as described in claim 1, characterized in that: Step S2 specifically includes the following processes: 1) Using the extreme learning machine method, the predicted value of stratigraphic lithology is obtained according to the following formula; T = H·β Where g(x) is the activation function, ω i For the input weights, b i Let T be the bias, β be the output matrix, β be the output weight, and H be the hidden layer output matrix, i = 1, 2, 3, ..., n; 2) Use the four-fold cross-validation method to determine the input weights ω and biases b of the model to be built, and then establish the lithology classification model.

5. The method for predicting drilling speed of small drilling rigs based on formation lithology classification as described in claim 1, characterized in that: Step S3 details The process includes the following: 1) The same extreme learning machine method is used to predict drilling speed; 2) Use the ten-fold cross-validation method to determine the input weight ω and bias b for each model to be built, and then build a drilling rate prediction model for each type of lithology; 3) Use the prediction results of the lithology classification model as the switching condition for the drilling speed prediction model, and complete the integration of the lithology classification model and the corresponding drilling speed prediction model. This allows drilling data at the same stage to be used as a judgment condition based on the lithology classification results, and to be specified as input to the corresponding lithology's drilling speed model for drilling speed prediction, thereby satisfying the requirement that data of the corresponding lithology can be input into the corresponding drilling speed model.

6. The method for predicting drilling speed of small drilling rigs based on formation lithology classification as described in claim 1, characterized in that: In step S5, during simulation verification, the calculation formula for the verification index of the test set is as follows: Where RMSE is the root mean square error, NRMSE is the normalized root mean square error, and y i It is measurement data. This is the predicted data, where n is the number of samples.

7. The method for predicting drilling speed of small drilling rigs based on formation lithology classification as described in claim 1, characterized in that: include: A processor and a storage device; the processor loads and executes the instructions and data stored in the storage device to implement the multi-model drilling speed prediction method for small drilling rigs based on stratigraphic lithology classification as described in any one of claims 1 to 6.

Images