Abnormality early warning method for rotary geosteering drilling

Abstract

The invention relates to the field of well drilling abnormity early warning, in particular to an abnormity early warning method for rotary geosteering well drilling, which comprises the following steps of: collecting historical well drilling data of a rotary geosteering tool, and screening out parameters related to abnormity of the rotary geosteering tool as a sensitive parameter combination; according to the method, a neural network early warning model with drilling data as an input layer is established, the neural network early warning model outputs an abnormal judgment result through input parameter data, and the neural network model is trained through combination of a plurality of groups of sensitive parameters, so that the neural network model has accurate judgment capability; the information input into the neural network model can be mapped into the abnormal type and the risk level, timely and accurate early warning is achieved, the safety risk of the rotary geosteering tool is reduced, drilling operation of the rotary geosteering tool is effectively guided, the rotary geosteering drilling abnormal condition is analyzed and judged in real time through drilling engineering parameters, the drilling efficiency is improved, and the drilling cost is reduced. And the economic loss is reduced.

Description

Technical Field

[0001] The present invention relates to the field of abnormal monitoring of drilling, and particularly to an abnormal early warning method for rotary geosteering drilling. Background Art

[0002] Existing marine shale gas generally takes the 1st - 4th sub - layers of the Longmaxi Formation as the main target layers. The well trajectory usually passes through the 2nd - 3rd sub - layers, and the thickness of its high - quality reservoir is 5 - 8m. In order to maximize the drilling encounter rate of the high - quality reservoir and ensure the smoothness of the wellbore trajectory, a rotary geosteering tool is designed for drilling operations. The rotary geosteering tool is not only expensive but also has high requirements for drilling working conditions. Once the rotary geosteering tool has malfunctions such as sticking or breaking, it may lead to situations such as wellbore collapse and spikes on the well path, affecting the efficiency, safety, and stability of drilling. The treatment of drilling anomalies is difficult, seriously affecting the efficiency and cost of the drilling project. Therefore, it is necessary to avoid abnormal situations in rotary geosteering drilling operations as much as possible.

[0003] However, in the current construction of rotary geosteering drilling operations, it is mainly the on - site engineers who judge whether there will be abnormal drilling states based on drilling parameters and their own experience. This requires high quality of on - site technicians, and the judgment results are greatly affected by subjectivity. As a result, there is no reliable means to monitor and early - warn abnormal situations during rotary geosteering drilling operations.

[0004] Therefore, there is an urgent need for a technical solution to solve the technical problem that there is no reliable means to monitor and early - warn abnormal situations during rotary geosteering drilling operations. Summary of the Invention

[0005] The purpose of the present invention is to provide an abnormal early warning method for rotary geosteering drilling to solve the technical problem that there is no reliable means to monitor and early - warn abnormal situations during rotary geosteering drilling operations.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows: An abnormal early warning method for rotary geosteering drilling, comprising the following steps: S1: Define the geological conditions of the work area: The work area includes the construction work area of the current rotary geosteering drilling. S2: Select early - warning parameters: Based on the geological conditions of the work area, initially screen out abnormal early - warning sample data from historical drilling data of similar or identical geological conditions. S3: Analyze the abnormal early - warning sample data: Conduct a sensitivity analysis on the abnormal early - warning sample data to determine the sensitive parameter combination for the current rotary geosteering drilling. S4: Establish a mathematical model: Establish a mathematical model by fitting the data of the sensitive parameter combinations. S5: Optimize the mathematical model: Obtain multiple sensitive parameter combinations by repeatedly performing steps S1 - S3, and optimize the algorithm of the mathematical model. S6: Conduct anomaly early warning: Input the drilling data of the current rotary steerable drilling into the mathematical model for calculation, and output the anomaly type and risk level from the mathematical model.

[0007] An anomaly early warning method for rotary steerable drilling according to the present invention analyzes in detail the geological conditions of the current rotary steerable drilling work area, compares the geological conditions of the current rotary steerable drilling with the historical drilling conditions, obtains the historical drilling data with the same geology as the current rotary steerable drilling from the historical drilling data, and preliminarily screens the historical data according to the geological conditions; then uses professional knowledge to screen the historical drilling data, filters out the data that has no correlation with the anomalies of rotary steerable drilling, and retains the parameters with obvious changes in parameter values as the anomaly early warning sample data; further conducts sensitivity analysis on the anomaly early warning sample data, eliminates the parameters with poor correlation with the anomalies of rotary steerable drilling, and retains the parameters with strong correlation with the anomalies of rotary steerable drilling as the parameters within the sensitive parameter combination. The selected sensitive parameters can more easily monitor drilling anomalies; a mathematical model is constructed using the sensitive parameter combination, and by repeatedly screening and performing sensitivity analysis on the historical drilling data, multiple sets of values of the sensitive parameter combination are obtained, and the algorithm of the mathematical model is optimized through the multiple sets of values to make the results output by the mathematical model more accurate; by inputting the sensitive parameters of the current rotary steerable drilling into the mathematical model for calculation, the anomaly type and risk level can be output from the mathematical model; by adopting an anomaly early warning method for rotary steerable drilling, the drilling process of the rotary steerable tool can be accurately controlled, and the operator can make early adjustments to the rotary steerable tool based on the early warning results obtained from the mathematical model.

[0008] As a preferred embodiment of the present invention, in S2, the historical drilling data includes the status parameters of the rotary geosteering tool and geological data. By comparing the parameter changes in the abnormal drilling state and the normal drilling state, the abnormal warning sample data is determined according to the amplitude of the parameter value change. Collecting the status parameters of the rotary geosteering tool and the geological parameters under the current status parameters, through the combination of multiple parameters, the accuracy of judging the abnormality of rotary geosteering drilling is improved; when selecting parameters from the historical drilling data, the parameters with large change amplitudes in the abnormal drilling state are selected as the abnormal warning sample data; the data of the selected abnormal warning sample parameters includes weight on bit, displacement, rotary speed, pump pressure, friction, torque, temperature, formation pressure, collapse pressure, drilling fluid density, drilling fluid viscosity, drilling fluid shear force, full angle change rate, rock type, rock drillability, rock compressive strength, rock hardness, faults, fractures and other data.

[0009] As a preferred embodiment of the present invention, in S3, the sensitivity analysis includes performing a correlation analysis on the abnormal warning sample data by using the clustering analysis method, and determining the parameters that make up the sensitive parameter combination according to the degree of parameter correlation. The clustering analysis method can more quickly select the parameter combination required for the sensitive parameter combination, improving the calculation efficiency; in some embodiments, through the correlation analysis method, the correlation analysis of the abnormal warning sample data can also be performed; the sensitive parameter combination after the sensitivity analysis is more sensitive to the abnormality of rotary geosteering drilling, and the risk level of drilling can be evaluated through the degree of numerical change of the parameters within the sensitive parameter combination.

[0010] As a preferred embodiment of the present invention, in S2, the abnormal warning sample data includes a stability combination, a safety combination and an efficiency combination; the stability combination is used to characterize the signal transmission state after the rotary geosteering tool enters the well; the safety combination is used to characterize the safety situation during the operation of the rotary geosteering tool; the efficiency combination is used to characterize the change in the drilling efficiency of the rotary geosteering tool and the smoothness of the drilling trajectory. During the construction process of rotary geosteering drilling, problems of stability, safety and efficiency will be encountered. Analyzing the problems of stability, safety and efficiency independently can more easily determine the risks encountered during the rotary geosteering drilling process, enabling the operators to more easily adjust the rotary geosteering tool for abnormal situations.

[0011] As a preferred embodiment of the present invention, in S3, by performing sensitivity analysis on the stability combination, the safety combination, and the efficiency combination respectively, sensitive parameter combinations within the stability combination, sensitive parameter combinations within the safety combination, and sensitive parameter combinations within the efficiency combination are obtained. The sensitive parameters of the stability combination obtained through cluster analysis include weight on bit, rotary speed, displacement, temperature, pressure, density, vibration intensity; the sensitive parameters of the safety combination include weight on bit, displacement, rotary speed, pump pressure, density, formation pressure, collapse pressure, dogleg severity, friction, torque, rock hardness, faults, fractures; the sensitive parameters of the efficiency combination include weight on bit, displacement, rotary speed, pump pressure, density, viscosity, shear force, friction, torque, dogleg severity, drillability of rock, rock hardness.

[0012] As a preferred embodiment of the present invention, in S4, a stability mathematical model, a safety mathematical model, and an efficiency mathematical model are respectively established. Different mathematical models perform calculations for different types of anomalies, and different risk levels of different types of anomalies can be output through different mathematical models; inputting the values of the stability-sensitive combination parameters into the stability mathematical model can output the risk level of stability anomalies; inputting the values of the safety-sensitive combination parameters into the safety mathematical model can output the risk level of safety anomalies; inputting the values of the efficiency-sensitive combination parameters into the efficiency mathematical model can output the risk level of efficiency anomalies.

[0013] As a preferred embodiment of the present invention, in S4, the mathematical models include a multiple linear regression model and a neural network model. Both the multiple linear regression and the neural network model can give early warnings for anomalies in rotary geosteering drilling; the multiple linear regression model can obtain calculation results based on the degree of change of sensitive parameters when an anomaly occurs; the neural network model can perform logical judgments on anomalies by simulating the human thinking mode.

[0014] As a preferred embodiment of the present invention, the expression of the multiple linear regression model is Y = α0 + α1X1 + α2X2 + … + α i X i , where: α0, α1, α2, …, α k are regression coefficients; X1, X2, …, X i are respectively the parameter values within the sensitive parameter combination. Through data fitting, the values of α0, α1, α2, …, α k as regression coefficients are determined. The sensitivity of each sensitive parameter to the anomalies in rotary geosteering drilling can be judged from the magnitudes of the regression coefficients. By substituting the values of the sensitive parameters into the multiple linear regression model and comparing the calculated values with different risk level thresholds, the risk level under the input drilling parameters can be determined.

[0015] As a preferred embodiment of the present invention, in S7, the neural network model includes an input layer, a hidden layer, and an output layer. The input layer is used to input the numerical values of sensitive parameters including x1, x2, …, x n and the output layer is used to output the risk levels including y1, y2, …, y m . The hidden layer can calculate the numerical values input by the input layer and send the calculation results to the output layer. The abnormal conditions of the rotary steerable tool are classified into risk levels of y1, y2, …, y m . A neural network model with the input information being the combination of sensitive parameters and the output information being the risk level is established. The neural network model is corrected by multiple groups of sensitive parameters for the model algorithm, and can accurately map the input information to the risk level. The hidden layer is an algorithm layer responsible for processing data. The abnormalities in rotary steerable drilling are usually classified into 3-5 risk levels. After the numerical values of the combination of sensitive parameters are input into the neural network model, the hidden layer can perform operations and processing on the data, so that the output layer of the neural network outputs the risk level.

[0016] As a preferred embodiment of the present invention, in S6, the rotary steerable drilling is monitored for alarms by inputting the real-time monitored drilling data into the mathematical model, and / or the risk of the rotary steerable drilling is predicted by inputting the predicted drilling data into the mathematical model. Through the mathematical model, the real-time monitoring of the rotary steerable drilling process can be realized, enabling the operator to adjust the rotary steerable tool in time through the alarm information and avoiding accidents caused by abnormal conditions during the drilling process. By inputting the predicted drilling data into the mathematical model, the risk during the drilling process can be better evaluated, the drilling risk can be better controlled, and safety accidents can be avoided.

[0017] In S5, the historical drilling data of different drilling projects are selected to obtain the combination of sensitive parameters. The historical drilling data of different drilling projects are more representative, enabling the obtained combination of sensitive parameters to be better trained and making the calculation results of the mathematical model more accurate.

[0018] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are: 1. An abnormal warning method for rotary geosteering drilling of the present invention analyzes in detail the geological conditions of the rotary geosteering drilling area this time, compares the geological conditions of the rotary geosteering drilling this time with the historical drilling conditions, obtains the historical drilling data with the same geology as the rotary geosteering drilling this time from the historical drilling data, and preliminarily screens the historical data according to the geological conditions. Subsequently, professional knowledge is used to screen the historical drilling data, filter out the data that has no correlation with the abnormalities of rotary geosteering drilling, and retain the parameters with obvious changes in parameter values as the abnormal warning sample data. Further, a sensitivity analysis is carried out on the abnormal warning sample data, excluding the parameters with poor correlation with the abnormalities of rotary geosteering drilling, and retaining the parameters with strong correlation with the abnormalities of rotary geosteering drilling as the parameters within the sensitive parameter combination. The selected sensitive parameters can more easily monitor the drilling abnormalities. A mathematical model is constructed using the sensitive parameter combination, and through repeated screening and sensitivity analysis of the historical drilling data, multiple sets of values of the sensitive parameter combination are obtained, and the algorithm of the mathematical model is optimized through multiple sets of values to make the results output by the mathematical model more accurate. By inputting the sensitive parameters of the rotary geosteering drilling this time into the mathematical model for calculation, the abnormal type and risk level can be output from the mathematical model. By adopting an abnormal warning method for rotary geosteering drilling, the drilling process of the rotary geosteering tool can be accurately controlled, and the operator can make early adjustments to the rotary geosteering tool based on the warning results obtained from the mathematical model.

[0019] 2. An abnormal warning method for rotary geosteering drilling of the present invention utilizes the obtained historical drilling data, screens out the sensitive parameter combination with strong correlation with the abnormalities of rotary geosteering drilling, establishes a mathematical model through the sensitive parameter combination, and optimizes the algorithm of the mathematical model through different sensitive parameter combinations to make the output result of the mathematical model more accurate, which can map the information input into the mathematical model to a certain risk level, thereby accurately warning the rotary geosteering drilling, reducing the safety risk of the rotary geosteering drilling operation, effectively guiding the rotary geosteering tool to carry out the drilling operation, clarifying the abnormal conditions during the drilling of the rotary geosteering tool through real-time analysis of the drilling engineering parameters, improving the drilling efficiency, reducing the economic loss, and having good economic value and practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a schematic flow chart of an abnormal warning method for rotary geosteering drilling; Figure 2 is a schematic flow chart of an abnormal warning method for rotary geosteering drilling in Embodiment 1; Figure 3 is a schematic flow chart of warning by the neural network warning model in Embodiment 1; Figure 4 It is a schematic diagram of the operation process of the neural network early warning model in Embodiment 1. Specific implementation manners

[0021] The present invention will be described in detail below with reference to the accompanying drawings.

[0022] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0023] Embodiment 1 As Figures 1 - 4 shown, an abnormal early warning method for rotary geosteering drilling, characterized by comprising the following specific steps: Step 1: Conduct geological exploration: obtain geological data of the drilling work area; conduct collection and analysis of geological materials: collect, sort out and analyze the construction geological materials of the current rotary geosteering drilling to clarify the geological conditions of the drilling work area; Step 2: Obtain historical drilling data: collect historical drilling data of rotary geosteering drilling operations; Step 3: Screen historical drilling data: select historical drilling data with the same geological conditions as data samples according to the geological conditions of the current drilling work area; Step 4: Select early warning parameters: in the historical drilling data with the same geology, conduct parameter screening through professional knowledge, and initially select parameters that can intuitively display the abnormalities of rotary geosteering tools as abnormal early warning sample data; Step 5: Form early warning parameter combinations: classify the parameters of the abnormal early warning samples into stability combinations, safety combinations and efficiency combinations; enable different early warning combinations to display different abnormal types of rotary geosteering drilling.

[0024] Step 6: Optimize sensitive parameter combinations: conduct sensitivity analysis on the abnormal early warning sample data to determine the sensitive parameter combinations for the abnormalities of rotary geosteering drilling; use the clustering analysis method to conduct correlation analysis on the parameters within the stability combination, safety combination and efficiency combination respectively, and respectively optimize the sensitive parameter combinations within the stability combination, safety combination and efficiency combination; Step 7: Establish an early warning model: use a neural network to perform data fitting on the sensitive parameters of the stability combination, the sensitive parameters of the safety combination and the sensitive parameters of the efficiency combination respectively, and establish a neural network early warning model for rotary geosteering drilling; Step 8: Perform model optimization: Repeatedly perform Steps S1 - S6 to obtain combinations of sensitive parameters in different historical drilling data. Train the neural network early warning model with multiple different combinations of sensitive parameters to obtain a neural network early warning model with more accurate calculations. Step 9: Conduct anomaly early warning: Input the drilling data into the neural network early warning model for calculation, and output the anomaly type and risk level from the neural network early warning model.

[0025] Specifically, in Step 2, the parameters in the historical drilling data include the state parameters obtained in real - time by the rotary steerable tool and the geological parameters during drilling. By comparing the numerical changes in parameters between the abnormal drilling state and the normal drilling state, initially select the parameters with relatively large numerical changes as the sample data for anomaly early warning; collect the state parameters of the rotary steerable tool and the geological parameters of the current state. Through the combination of multiple parameters, improve the accuracy of judging the anomalies of the rotary steerable tool; when selecting the parameters in the historical drilling data, select the parameters with relatively large changes in the abnormal drilling state as the sample data for anomaly early warning; the selected sample data for anomaly early warning includes data such as weight on bit, flow rate, rotary speed, pump pressure, friction, torque, temperature, formation pressure, collapse pressure, drilling fluid density, drilling fluid viscosity, drilling fluid shear force, total angle change rate, rock type, rock drillability, rock compressive strength, rock hardness, faults, fractures, etc.

[0026] Specifically, in Step 5, the stability combination is used to ensure normal signal transmission after the rotary steerable tool is lowered into the well, and ensure that the rotary steerable tool can be stably controlled during operation; the safety combination is used to ensure the safety of the rotary steerable tool during operation, and ensure that the tool does not occur complex situations such as stabbing, jamming, and breaking under corresponding drilling parameter conditions; the efficiency combination is used to ensure the drilling efficiency of the rotary steerable tool and the smoothness of the drilling trajectory, and guarantee the drilling efficiency.

[0027] Specifically, in Step 6, after screening the combinations of sensitive parameters through the clustering analysis method, the combinations of sensitive parameters for the stability combination include weight on bit, rotary speed, flow rate, temperature, pressure, density, vibration intensity; the combinations of sensitive parameters for the safety combination include weight on bit, flow rate, rotary speed, pump pressure, density, formation pressure, collapse pressure, total angle change rate, friction, torque, rock hardness, faults, fractures; the combinations of sensitive parameters for the efficiency combination include weight on bit, flow rate, rotary speed, pump pressure, density, viscosity, shear force, friction, torque, total angle change rate, rock drillability, rock hardness.

[0028] Specifically, in step 7, a stability warning model, a safety warning model, and an efficiency warning model are established in the form of a neural network respectively; different abnormal types calculate the risk level through different neural network models. After the drilling data is input into the stability warning model, the safety warning model, and the efficiency warning model, the stability warning model can output the stability risk level, the safety warning model can output the safety risk level, and the efficiency warning model can output the efficiency risk level, so as to process the input drilling data.

[0029] Specifically, in step 8, historical drilling data of different drilling projects are selected to obtain sensitive parameter combinations; the historical drilling data of different drilling projects are more representative, enabling the obtained sensitive parameter combinations to be better trained and making the operation results of the neural network warning model more accurate.

[0030] Specifically, taking the stability warning model as an example, the risk levels y1, y2, and y3 are set as low risk, medium risk, and high risk respectively, and each risk level is set with a threshold. For example, the low risk is 100 - 200, the medium risk is 200 - 300, and the high risk is greater than 300. x1 - x7 are the values of weight on bit, rotary speed, displacement, temperature, pressure, density, and vibration intensity respectively; the specific values of x1 - x 7， are input into the stability warning model. After the input values are processed by the hidden layer through operations, an intermediate value of 249 is obtained. Comparing this intermediate value with the threshold, since the intermediate value is within 200 - 300, it is determined that the risk level output by the output layer is y2 medium risk, thus realizing the stability warning of the rotary steerable tool; if the intermediate value is less than 100, no risk level is output.

[0031] Specifically, in step 9, the rotary steerable tool is monitored for alarms through the drilling data real - time monitored at the input layer of the neural network warning model.

[0032] Preferably, when the abnormal conditions of the rotary steerable tool are divided into 5 risk levels, namely low risk, medium - low risk, medium risk, medium - high risk, and high risk, different thresholds are set for different risk levels respectively.

[0033] Furthermore, to improve the operation efficiency, a data screening module is also set in the neural network warning model, so that the drilling parameter data is first screened after entering the neural network warning model, and the parameter types entering the stability warning model, the safety warning model, and the efficiency warning model for operation are the sensitive parameters of the stability warning model, the safety warning model, and the efficiency warning model respectively, thus improving the operation efficiency of the neural network warning model.

[0034] An abnormal warning method for rotary geosteering drilling in this embodiment screens historical drilling data, establishes a neural network warning model, and trains the neural network warning model with a large amount of data to make the judgment of the neural network warning model more accurate; the drilling data can be input into the neural network warning model, and after being calculated by the neural network warning model, the abnormal type and risk level are output in real time to monitor the rotary geosteering drilling in real time.

[0035] An abnormal warning method for rotary geosteering drilling in the present invention analyzes in detail the geological conditions of the present rotary geosteering drilling, compares the geological conditions of this drilling run with historical drilling conditions, obtains historical drilling data with the same geology as the present rotary geosteering drilling from the historical drilling data, and preliminarily screens the historical data according to the geological conditions; then, professionals use professional knowledge to screen the historical drilling data, filter out the data that has no relevance to the abnormal conditions of the rotary geosteering tool, and initially select the parameters that have relevance to the abnormal conditions of the rotary geosteering tool as the abnormal warning sample data; further conduct a sensitivity analysis on the abnormal warning sample data, eliminate the parameters with poor relevance to the abnormal conditions of the rotary geosteering drilling, and retain the parameters with strong relevance to the abnormal conditions of the rotary geosteering drilling as the parameters within the sensitive parameter combination. The selected sensitive parameters can more easily detect drilling abnormalities; use the sensitive parameter combination to construct a mathematical model, and through repeatedly screening and conducting sensitivity analysis on the historical drilling data, obtain the numerical values of multiple groups of sensitive parameter combinations, and optimize the algorithm of the mathematical model through multiple groups of sensitive parameter combinations to make the results output by the mathematical model more stable; input the numerical values of the sensitive parameters into the obtained mathematical model, and output the abnormal type and risk level from the mathematical model; by adopting an abnormal warning method for rotary geosteering drilling, the drilling process of the rotary geosteering tool can be accurately controlled, and the operator can make early adjustments to the rotary geosteering tool based on the warning results obtained from the mathematical model.

[0036] Embodiment 2 An abnormal warning method for rotary geosteering drilling in this embodiment has an implementation manner that is substantially the same as that of Embodiment 1. The difference from Embodiment 1 is that in S9, risk prediction is performed on the rotary geosteering tool by inputting the predicted drilling data into the neural network warning model.

[0037] An abnormal warning method for rotary geosteering drilling in this embodiment enables the neural network warning model to output the operation results based on the predicted drilling data by inputting the predicted drilling data, so that the neural network warning model can be used for the drilling prediction of the rotary geosteering tool.

[0038] Embodiment 3 An abnormal warning method for rotary geosteering drilling in this embodiment is implemented in a manner substantially the same as that in Embodiment 1. Different from Embodiment 1, the established mathematical model is a multiple linear regression model.

[0039] Specifically, the expression of the multiple linear regression model is Y = α0 + α1X1 + α2X2 + … + α i X i , where: α0, α1, α2, …, α k are regression coefficients; X1, X2, …, X i are the numerical values of different parameters respectively, and different risk level thresholds can be determined according to the simulation results.

[0040] Specifically, establish a stability multiple model, a safety multiple model, and an efficiency multiple model; by inputting the parameter values of weight on bit, rotary speed, displacement, temperature, pressure, density, vibration intensity into the stability multiple model, the risk level of stability abnormality can be determined; by inputting the parameter values of weight on bit, displacement, rotary speed, pump pressure, density, formation pressure, collapse pressure, build rate, friction, torque, rock hardness, fault, fracture into the safety multiple model, the risk level of safety abnormality can be determined; by inputting the parameter values of weight on bit, displacement, rotary speed, pump pressure, density, viscosity, shear force, friction, torque, build rate, drillability of rock, rock hardness into the efficiency multiple model, the risk level of efficiency abnormality can be determined.

[0041] Specifically, the regression coefficients α0, α1, α2, …, α k represent the weight magnitudes of sensitive parameters. The larger the regression coefficient, the greater the influence of the corresponding parameter on the corresponding abnormality.

[0042] An abnormal warning method for rotary geosteering drilling in this embodiment determines the numerical values of α0, α1, α2, …, α k as the regression coefficients through data fitting. The sensitivity degrees of each sensitive parameter to the abnormality of rotary geosteering drilling can be judged from the magnitudes of the regression coefficients. Substitute the values of the sensitive parameters into the multiple linear regression model, and compare the values obtained through calculation with different risk level thresholds, so as to determine the risk level of drilling abnormality under the input drilling parameters.

[0043] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An abnormal warning method for rotary geosteering drilling, characterized in that, It includes the following steps: S1: Define the geological conditions of the work area: The work area includes the construction work area for the current rotary geosteering drilling; S2: Select warning parameters: Based on the geological conditions of the work area, initially screen out abnormal warning sample data from the historical drilling data of similar or identical geological conditions; S3: Analyze the abnormal warning sample data: Conduct a sensitivity analysis on the abnormal warning sample data to determine the sensitive parameter combination for the current rotary geosteering drilling; S4: Establish a mathematical model: Establish a mathematical model by fitting the data of the sensitive parameter combination; S5: Optimize the mathematical model: By repeatedly performing steps S1 - S3, obtain multiple sensitive parameter combinations and optimize the algorithm of the mathematical model; S6: Conduct abnormal warning: Input the drilling data of the current rotary geosteering drilling into the mathematical model for calculation, and output the abnormal type and risk level from the mathematical model.

2. The abnormal warning method for rotary geosteering drilling according to claim 1, wherein In S2, the historical drilling data includes the status parameters and geological data of the rotary geosteering tool. Compare the parameter changes in the abnormal drilling state and the normal drilling state, and determine the abnormal warning sample data according to the amplitude of the parameter value change.

3. The abnormal warning method for rotary geosteering drilling according to claim 1, characterized in that, In S3, the sensitivity analysis includes using the clustering analysis method to conduct a correlation analysis on the abnormal warning sample data, and determining the parameters that make up the sensitive parameter combination according to the parameter correlation degree.

4. The abnormal warning method for rotary geosteering drilling according to claim 1, wherein In S2, the abnormal warning sample data includes a stability combination, a safety combination, and an efficiency combination; the stability combination is used to characterize the signal transmission state after the rotary geosteering tool enters the well; the safety combination is used to characterize the safety situation during the operation of the rotary geosteering tool; the efficiency combination is used to characterize the change in the drilling efficiency of the rotary geosteering tool and the smoothness of the drilling trajectory.

5. The abnormal warning method for rotary geosteering drilling according to claim 4, wherein In S3, by conducting sensitivity analysis on the stability combination, the safety combination, and the efficiency combination respectively, obtain the sensitive parameter combination within the stability combination, the sensitive parameter combination within the safety combination, and the sensitive parameter combination within the efficiency combination.

6. The abnormal warning method for rotary geosteering drilling according to claim 5, characterized in that, In S4, establish a stability mathematical model, a safety mathematical model, and an efficiency mathematical model respectively.

7. The abnormal warning method for rotary geosteering drilling according to claim 1, wherein In S4, the mathematical model includes a multiple linear regression model and a neural network model.

8. The abnormal warning method for rotary geosteering drilling according to claim 7, characterized in that, The expression of the multiple linear regression model is Y = α0 + α1X1 + α2X2 + … + α i X i , where: α0, α1, α2, …, α k are regression coefficients; X1, X2, …, X i are the parameter values within the sensitive parameter combination respectively.

9. The abnormal warning method for rotary geosteering drilling according to claim 7, characterized in that, The neural network model includes an input layer, a hidden layer, and an output layer. The input layer is used to input the numerical values of sensitive parameters including x1, x2, …, x n and the output layer is used to output the risk levels including y1, y2, …, y m . The hidden layer can calculate the values input by the input layer and send the calculation results to the output layer.

10. An abnormal warning method for rotary geosteering drilling according to claim 1, characterized in that, In S6, conduct alarm monitoring on the rotary geosteering drilling by inputting the real-time monitored drilling data into the mathematical model, and / or conduct risk prediction on the rotary geosteering drilling by inputting the predicted drilling data into the mathematical model.

Images