An electrically controlled well drilling safety release control method and system

Abstract

The application discloses an electrically-controlled well drilling safety release control method and system, and relates to the field of intelligent control, wherein the method comprises the following steps: collecting environment information sets and operation information sets, and obtaining detection data sets; constructing a release control analysis model; inputting the detection data sets into Q comprehensive analysis units in the release control analysis model to obtain Q first analysis results, and inputting the environment information sets and the operation information sets into Q dimension analysis units in the release control analysis model to obtain Q second analysis results; obtaining a comprehensive abnormal probability value and a comprehensive normal probability value, and controlling a target release structure to perform a release operation when the comprehensive abnormal probability value is greater than a preset abnormal threshold. The technical problems that the control precision of the electrically-controlled well drilling safety release is insufficient in the prior art, and the control effect of the electrically-controlled well drilling safety release is poor are solved. The technical effect of improving the control quality of the electrically-controlled well drilling safety release is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent control, in particular, to an electric control well drilling safety release control method and system. BACKGROUND

[0002] The well drilling safety release is an important well drilling operation tool. When the drilling equipment is stuck in the well due to an accident, the electric control well drilling safety release can be started in time to separate the lower pipe string of the drilling equipment and lift the upper pipe string to the well head, so as to separate the upper and lower parts of the drilling equipment and realize effective protection of the drilling equipment.

[0003] In the prior art, there are technical problems of poor control accuracy of the electric control well drilling safety release, low separation success rate of the electric control well drilling safety release, and high possibility of misoperation of the electric control well drilling safety release, which leads to poor control effect of the electric control well drilling safety release. SUMMARY

[0004] The present application provides an electric control well drilling safety release control method and system. The technical problems of poor control accuracy of the electric control well drilling safety release, low separation success rate of the electric control well drilling safety release, and high possibility of misoperation of the electric control well drilling safety release, which leads to poor control effect of the electric control well drilling safety release, are solved. The technical effects of improving the control accuracy of the electric control well drilling safety release, improving the separation success rate of the electric control well drilling safety release, avoiding the misoperation of the electric control well drilling safety release, and improving the control quality of the electric control well drilling safety release are achieved.

[0005] In view of the above problems, the present application provides an electric control well drilling safety release control method and system.

[0006] In a first aspect, the application provides an electrically controlled well drilling safety release control method, wherein the method is applied to an electrically controlled well drilling safety release control system, and the method comprises: in the process of drilling by using a target well drilling device, detecting P kinds of environmental information in a well and R kinds of running information in the target well drilling device by Q detection devices arranged on the target well drilling device, Q, P and R are integers greater than 1, Q is the sum of P and R, the P kinds of environmental information include at least one of a first pressure, a second pressure and a temperature, the R kinds of running information include at least one of a displacement per unit time, a vibration frequency, a vibration peak value and an acceleration, and the target well drilling device includes a target release structure; collecting environmental information sets and running information sets, and obtaining detection data sets; based on drilling detection data in a historical time, constructing a release control analysis model, wherein the release control analysis model is constructed based on ensemble learning and includes Q comprehensive analysis units and Q dimension analysis units; inputting the detection data sets into the Q comprehensive analysis units to obtain Q first analysis results, and inputting Q data in the environmental information sets and the running information sets into the Q dimension analysis units respectively to obtain Q second analysis results; performing weighted calculation on the Q first analysis results and the Q second analysis results to obtain a comprehensive abnormal probability value and a comprehensive normal probability value, and when the comprehensive abnormal probability value is greater than a preset abnormal threshold, controlling the target release structure to perform a release operation.

[0007] In a second aspect, the application also provides an electrically controlled well drilling safety release control system, wherein the system comprises: a well drilling detection module, configured to detect P types of environmental information in a well and R types of operation information in a target well drilling device by Q detection devices arranged on the target well drilling device during well drilling by the target well drilling device, Q, P and R are integers greater than 1, Q is the sum of P and R, the P types of environmental information include at least one of a first pressure, a second pressure and a temperature, and the R types of operation information include at least one of a displacement per unit time, a vibration frequency, a vibration peak value and an acceleration, and the target well drilling device comprises a target release structure; a detection data set obtaining module, configured to collect environmental information sets and operation information sets and obtain a detection data set; a construction module, configured to construct a release control analysis model based on well drilling detection data in a historical time, wherein the release control analysis model is constructed based on ensemble learning and comprises Q comprehensive analysis units and Q dimension analysis units; a comprehensive analysis module, configured to input the detection data set into the Q comprehensive analysis units to obtain Q first analysis results, and input Q data in the environmental information sets and the operation information sets into the Q dimension analysis units to obtain Q second analysis results; and a release control module, configured to perform weighted calculation on the Q first analysis results and the Q second analysis results to obtain a comprehensive abnormal probability value and a comprehensive normal probability value, and control the target release structure to perform a release operation when the comprehensive abnormal probability value is greater than a preset abnormal threshold.

[0008] The one or more technical solutions provided in the application have at least the following technical effects or advantages:

[0009] During well drilling by a target well drilling device, Q detection devices arranged on the target well drilling device are used to obtain environmental information sets, operation information sets and a detection data set; a release control analysis model is constructed based on well drilling detection data in a historical time; the detection data set is input into Q comprehensive analysis units of the release control analysis model to obtain Q first analysis results, and the environmental information sets and the operation information sets are input into Q dimension analysis units of the release control analysis model to obtain Q second analysis results; weighted calculation is performed on the Q first analysis results and the Q second analysis results to obtain a comprehensive abnormal probability value and a comprehensive normal probability value, and a target release structure is controlled to perform a release operation when the comprehensive abnormal probability value is greater than a preset abnormal threshold. The technical effects of improving the control accuracy of electrically controlled well drilling safety release, improving the disengagement success rate of electrically controlled well drilling safety release, avoiding misoperation of electrically controlled well drilling safety release and improving the control quality of electrically controlled well drilling safety release are achieved.

[0010] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the following specific embodiments of the present application can be implemented according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application will be described in detail. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced. Obviously, the drawings described below only relate to some embodiments of the present disclosure, not to limit the present disclosure.

[0012] Figure 1 Flowchart of a safety release control method for electrically controlled drilling of the present application;

[0013] Figure 2 Flowchart of obtaining a detection data set in a safety release control method for electrically controlled drilling of the present application;

[0014] Figure 3 Structure diagram of a safety release control system for electrically controlled drilling of the present application.

[0015] Explanation of reference signs: drilling detection module 11, detection data set obtaining module 12, construction module 13, comprehensive analysis module 14, release control module 15. DETAILED DESCRIPTION

[0016] The present application provides a safety release control method and system for electrically controlled drilling. The technical problems of poor control accuracy, low success rate of release, and high possibility of misoperation of the safety release control of electrically controlled drilling in the prior art, which leads to poor control effect of the safety release control of electrically controlled drilling, are solved. The technical effects of improving the control accuracy of the safety release control of electrically controlled drilling, improving the success rate of release of the safety release control of electrically controlled drilling, avoiding misoperation of the safety release control of electrically controlled drilling, and improving the control quality of the safety release control of electrically controlled drilling are achieved.

[0017] Embodiment one

[0018] Please refer to the accompanying Figure 1 The present application provides a safety release control method for electrically controlled drilling, wherein the method is applied to a safety release control system for electrically controlled drilling, and the method specifically comprises the following steps:

[0019] Step S100: In the process of drilling by using a target drilling equipment, P kinds of environmental information in the well and R kinds of running information in the target drilling equipment are detected by Q detection devices arranged on the target drilling equipment, Q, P and R are integers greater than 1, Q is the sum of P and R, the P kinds of environmental information include at least one of first pressure, second pressure and temperature, and the R kinds of running information include at least one of displacement per unit time, vibration frequency, vibration peak value and acceleration, and the target drilling equipment includes a target releasing structure;

[0020] Specifically, in the process of drilling by using a target drilling equipment, P kinds of environmental information in the well and R kinds of running information in the target drilling equipment are detected by Q detection devices. The target drilling equipment can be any drilling equipment that uses the electrically controlled drilling safety releasing control system for intelligent releasing control. The target drilling equipment includes a target releasing structure. The target releasing structure is the electrically controlled drilling safety releasing of the target drilling equipment. When the target drilling equipment is stuck in an accident in the well, the target releasing structure can be started in time to effectively protect the target drilling equipment. The Q detection devices are in communication connection with the target drilling equipment. The Q detection devices include pressure sensors, temperature sensors, displacement sensors, vibration sensors and acceleration sensors in the prior art. The P kinds of environmental information include first pressure, second pressure and temperature. The first pressure is the downward drilling working pressure received by the target drilling equipment. The second pressure is the well wall pressure received by the target drilling equipment in drilling. The temperature is the drilling environmental temperature received by the target drilling equipment. The R kinds of running information include displacement per unit time, vibration frequency, vibration peak value and acceleration of the target drilling equipment. Q, P and R are all integers greater than 1, and Q = P + R.

[0021] Step S200: Collecting the environmental information set and the running information set, and obtaining a detection data set;

[0022] Further, as shown in the accompanying drawings, Figure 2 the step S200 of the present application further includes:

[0023] Step S210: In a plurality of continuous time windows according to a preset frequency, the P kinds of environmental information and the R kinds of running information are detected by the Q detection devices to obtain a plurality of initial environmental information sets and a plurality of initial running information sets;

[0024] Step S220: The data of the same kind of environmental information and the data of the same kind of running information in the plurality of initial environmental information sets and the plurality of initial running information sets are respectively subjected to mean value calculation to obtain the environmental information set and the running information set;

[0025] Step S230: The environmental information set and the running information set are integrated to obtain the detection data set.

[0026] Specifically, when drilling is performed by using the target drilling equipment, the target drilling equipment is monitored in real time by the Q detection devices according to a plurality of continuous time windows based on a preset frequency, and a plurality of initial environment information sets and a plurality of initial operation information sets are obtained. The preset frequency includes a drilling equipment monitoring frequency that is set in advance. The plurality of time windows include a plurality of continuous time points corresponding to the preset frequency. Each initial environment information set includes real-time first pressure parameters, real-time second pressure parameters, and real-time temperature parameters corresponding to P kinds of environment information in each time window. Each initial operation information set includes real-time unit time displacement, real-time vibration frequency, real-time vibration peak value, and real-time acceleration of the target drilling equipment corresponding to R kinds of operation information in each time window.

[0027] Further, the plurality of initial environment information sets are subjected to cluster analysis based on the P kinds of environment information, a plurality of data corresponding to the same kind of environment information in the plurality of initial environment information sets are classified into one category, a plurality of sets of clustered environment information are obtained, and mean calculation is performed on the plurality of sets of clustered environment information to obtain an operation information set. Each set of clustered environment information includes a plurality of data corresponding to the same kind of environment information in the plurality of initial environment information sets. That is, the plurality of sets of clustered environment information include first pressure clustered environment information, second pressure clustered environment information, and temperature clustered environment information corresponding to the P kinds of environment information. The environment information set includes a first pressure average, a second pressure average, and a temperature average corresponding to the plurality of sets of clustered environment information.

[0028] Similarly, the plurality of initial operation information sets are subjected to cluster analysis and mean calculation based on the R kinds of operation information, and an operation information set is obtained, which, in combination with the environment information set, obtains a detection data set. The operation information set includes a unit time displacement average, a vibration frequency average, a vibration peak value average, and an acceleration average corresponding to the plurality of initial operation information sets. The operation information set and the environment information set are calculated in the same way, and for the sake of brevity of the description, details are not repeated here. The detection data set includes the environment information set and the operation information set.

[0029] The technical effect of improving the reliability of the safety release control of the target drilling equipment is achieved by monitoring the target drilling equipment in real time by the Q detection devices to obtain comprehensive and accurate detection data sets.

[0030] Step S300: based on drilling detection data in a historical time, a release control analysis model is constructed, wherein the release control analysis model is constructed based on ensemble learning and includes Q comprehensive analysis units and Q dimension analysis units.

[0031] Further, the step S300 of the present application further includes:

[0032] Step S310: based on the drilling detection data in the historical time, obtain a plurality of sample detection data sets, a plurality of sample environment information sets, a plurality of sample operation information sets and a sample abnormality judgment result set, the sample abnormality judgment result includes the judgment result of whether the drilling operation is abnormal and needs to be performed.

[0033] Specifically, the one kind electric control drilling safety release control system is connected, the drilling detection data in the historical time is collected for the one kind electric control drilling safety release control system, a plurality of sample detection data sets, a plurality of sample environment information sets, a plurality of sample operation information sets and a sample abnormality judgment result set are obtained.

[0034] The historical time includes a plurality of historical time intervals. Each group of historical time intervals includes a plurality of historical time windows. Each sample environment information set includes a historical first pressure average value, a historical second pressure average value and a historical temperature average value corresponding to each group of historical time intervals. Each sample operation information set includes a historical unit time displacement average value, a historical vibration frequency average value, a historical vibration peak average value and a historical acceleration average value corresponding to each group of historical time intervals. Each sample detection data set includes a sample environment information set and a sample operation information set corresponding to the same group of historical time intervals. The sample abnormality judgment result set includes a plurality of sample abnormality judgment results corresponding to a plurality of sample detection data sets. Each sample abnormality judgment result includes the judgment result of whether the drilling operation is abnormal and needs to be performed, that is, each sample abnormality judgment result includes whether the drilling operation corresponding to each sample detection data set is abnormal, and whether the drilling operation corresponding to the sample detection data set needs to be performed.

[0035] Step S320: using the plurality of sample detection data sets and the sample abnormality judgment result set as construction data, constructing the Q comprehensive analysis units;

[0036] Further, step S320 of the present application further includes:

[0037] Step S321: dividing and combining the data in the plurality of sample detection data sets and the sample abnormality judgment result set to obtain K groups of construction data;

[0038] Step S322: assigning K first comprehensive weights to the K groups of construction data, each first comprehensive weight being 1 / K;

[0039] Step S323: using the K groups of construction data, based on the BP neural network, constructing a first comprehensive analysis unit in the Q comprehensive analysis units obtained by training;

[0040] Specifically, based on the plurality of sample detection data sets and the sample anomaly judgment result set, data division and combination are performed to obtain K sets of construction data. Each set of construction data includes any one sample detection data set and the sample anomaly judgment result corresponding to the sample detection data set.

[0041] Further, the K sets of construction data are assigned with weights to obtain K first comprehensive weights corresponding to the K sets of construction data. Based on the K first comprehensive weights, cross-supervision training of the K sets of construction data is performed by using a BP neural network to obtain a first comprehensive analysis unit, and the first comprehensive analysis unit is added to the Q comprehensive analysis units. Each first comprehensive weight is 1 / K. The BP neural network is a multi-layer feedforward neural network trained according to an error back propagation algorithm. The BP neural network includes an input layer, a plurality of layers of neurons, and an output layer. The BP neural network can perform forward calculation and reverse calculation. During forward calculation, input information is processed layer by layer from the input layer through the plurality of layers of neurons, and is transferred to the output layer. The state of each layer of neurons only affects the state of the next layer of neurons. If the desired output cannot be obtained at the output layer, reverse calculation is performed, the error signal is returned along the original connection path, and the error signal is minimized by modifying the weights of the neurons. The first comprehensive analysis unit includes an input layer, a hidden layer, and an output layer.

[0042] Step S324: The K sets of construction data are used to test the first comprehensive accuracy of the first comprehensive analysis unit, and K second comprehensive weights are calculated based on the K first comprehensive weights.

[0043] The K second comprehensive weights are calculated based on the K first comprehensive weights and the first comprehensive accuracy of the first comprehensive analysis unit, as follows:

[0044]

[0045] Wherein, W2(B1) is the second comprehensive weight of the first set of construction data B1, W1(B1) is the first comprehensive weight of the first set of construction data, Z1 is the first accuracy of the first comprehensive analysis unit, T is the number of currently constructed comprehensive analysis units, T=1, 2, 3…Q.

[0046] Step S325: Based on the BP neural network, the K second comprehensive weights are used to construct and train a second comprehensive analysis unit in the Q comprehensive analysis units, wherein the training computing resource of each set of construction data is positively correlated with the size of the second comprehensive weight.

[0047] Step S326: The Q comprehensive analysis units are continuously constructed.

[0048] Specifically, the K sets of construction data are input as input information into the first comprehensive analysis unit to obtain K predicted anomaly judgment results corresponding to the K sets of construction data. The accuracy rate is calculated based on the K predicted anomaly judgment results and K sample anomaly judgment results in the K sets of construction data to obtain a first comprehensive accuracy rate of the first comprehensive analysis unit.

[0049] For example, when the accuracy rate is calculated based on the K predicted anomaly judgment results and the K sample anomaly judgment results in the K sets of construction data, the K predicted anomaly judgment results are compared with the corresponding K sample anomaly judgment results based on the K sets of construction data. When the predicted anomaly judgment result is consistent with the corresponding sample anomaly judgment result, the predicted anomaly judgment result is marked as an accurate predicted anomaly judgment result. The number of accurate predicted anomaly judgment results in the K predicted anomaly judgment results is set as the number of accurate judgment results. The ratio between the number of accurate judgment results and K is output as the first comprehensive accuracy rate.

[0050] Further, the weight value is adjusted based on the first comprehensive accuracy rate and the K first comprehensive weight values to obtain K second comprehensive weight values.

[0051] For example, the weight value is adjusted based on the first comprehensive accuracy rate and the K first comprehensive weight values as follows:

[0052]

[0053] wherein each set of construction data in the K sets of construction data is sequentially set as a first set of construction data B1, W2(B1) is the second comprehensive weight value of the output first set of construction data B1, W1(B1) is the first comprehensive weight value of the input first set of construction data, Z1 is the first accuracy rate of the input first comprehensive analysis unit, T is the number of currently constructed comprehensive analysis units, and T=1, 2, 3…Q.

[0054] Further, the K sets of construction data are matched with training computing resources according to the K second comprehensive weight values to obtain training computing resources corresponding to each set of construction data in the K sets of construction data. The K sets of construction data are supervised trained through a BP neural network based on the training computing resources corresponding to each set of construction data to obtain a second comprehensive analysis unit. Then, Q comprehensive analysis units are continuously constructed based on the K sets of construction data. The training computing resources include the training times corresponding to each set of construction data. The training computing resources of each set of construction data are positively correlated with the size of the second comprehensive weight value. The greater the second comprehensive weight value, the more the training times of the construction data corresponding to the second comprehensive weight value when the second comprehensive analysis unit is constructed. The Q comprehensive analysis units are continuously constructed based on the K sets of construction data in the same way as the first comprehensive analysis unit and the second comprehensive analysis unit, which is not described herein for the sake of brevity.

[0055] The technical effect of improving the accuracy of the trip control analysis model is achieved by training a plurality of sample detection data sets and sample anomaly judgment result sets to construct Q comprehensive analysis units with high accuracy and generalization performance.

[0056] Step S330: using the plurality of sample environment information sets, the plurality of sample running information sets, and the sample anomaly judgment result set as construction data, constructing the Q-dimensional analysis units;

[0057] Further, step S330 of the present application further comprises:

[0058] Step S331: using the sample environment information of the first environment information of the P kinds of environment information in the K sets of construction data and the K sample anomaly judgment results as K sets of first-dimensional construction data, constructing a first-dimensional analysis unit in the Q-dimensional analysis units based on the BP neural network and training;

[0059] Step S332: using the K sets of first-dimensional construction data to test the first-dimensional analysis unit to obtain a first-dimensional accuracy rate;

[0060] Step S333: calculating K first-dimensional weights according to the first-dimensional accuracy rate and the K first comprehensive weights;

[0061] Step S334: using the sample environment information of the second environment information of the P kinds of environment information in the K sets of construction data and the K sample anomaly judgment results as K sets of second-dimensional construction data, constructing a second-dimensional analysis unit in the Q-dimensional analysis units based on the BP neural network and training according to the K first-dimensional weights, wherein the training computing resource of each set of second-dimensional construction data is positively correlated with the size of the first-dimensional weight;

[0062] Step S335: using the K sets of second-dimensional construction data to test the second-dimensional analysis unit to obtain a second-dimensional accuracy rate, calculating K second-dimensional weights, and continuing to construct the Q-dimensional analysis units.

[0063] Step S340: integrating the Q comprehensive analysis units and the Q-dimensional analysis units to obtain the trip control analysis model.

[0064] Specifically, based on P kinds of environmental information, a first environmental information is obtained by random selection. Based on the first environmental information, K groups of first dimension construction data are obtained by data extraction on the K groups of construction data. Then, based on K first comprehensive weights, the K groups of first dimension construction data are continuously self-trained and learned to a convergence state by a BP neural network to obtain a first dimension analysis unit, and the first dimension analysis unit is added to Q dimension analysis units. Each group of first dimension construction data includes sample environmental information corresponding to the first environmental information and a corresponding sample abnormality judgment result in each group of construction data of the K groups of construction data.

[0065] The first dimension analysis unit includes an input layer, a hidden layer, and an output layer.

[0066] Further, the K groups of first dimension construction data are input into the first dimension analysis unit as input information to obtain K predicted abnormality judgment results corresponding to the K groups of first dimension construction data, and a first dimension accuracy is obtained by accuracy calculation combined with K sample abnormality judgment results in the K groups of first dimension construction data. Then, based on the first dimension accuracy and the K first comprehensive weights, K first dimension weights are obtained. The first dimension accuracy is calculated in the same way as the first comprehensive accuracy, and for the sake of brevity of the specification, it will not be repeated here. The calculation method of the K first dimension weights is the same as that of the K second comprehensive weights, and for the sake of brevity of the specification, it will not be repeated here.

[0067] Further, the P kinds of environmental information are randomly selected again to obtain a second environmental information. Based on the second environmental information, K groups of second dimension construction data are obtained by data extraction on the K groups of construction data. Then, based on the K first dimension weights, the K groups of second dimension construction data are trained and matched with training computing resources to obtain training computing resources of each group of second dimension construction data. Further, based on the training computing resources of each group of second dimension construction data, the K groups of second dimension construction data are cross-supervised trained by a BP neural network to obtain a second dimension analysis unit, and the second dimension analysis unit is added to the Q dimension analysis units. Each group of second dimension construction data includes sample environmental information corresponding to the second environmental information and a sample abnormality judgment result corresponding to the sample environmental information in each group of construction data of the K groups of construction data. The training computing resources of each group of second dimension construction data include a training number corresponding to each group of second dimension construction data. The training computing resources of each group of second dimension construction data are positively correlated with the first dimension weight. The larger the first dimension weight, the more the training number of the second dimension construction data corresponding to the first dimension weight when constructing the second dimension analysis unit.

[0068] Furthermore, the K sets of second-dimensional construction data are input into the second-dimensional analysis unit to obtain the second-dimensional accuracy. K second-dimensional weights are calculated based on the second-dimensional accuracy, and Q dimension analysis units are then used based on these K second-dimensional weights. Combined with Q comprehensive analysis units, a drop-out control analysis model is obtained. The calculation method for the second-dimensional accuracy is the same as that for the first comprehensive accuracy, and will not be repeated here for the sake of brevity. The calculation method for the K second-dimensional weights is the same as that for the sake of brevity, will not be repeated here. The construction method for the Q dimension analysis units is the same as that for the first and second dimension analysis units, and will not be repeated here for the sake of brevity. The drop-out control analysis model includes Q comprehensive analysis units and Q dimension analysis units. Each comprehensive analysis unit has a corresponding comprehensive accuracy, and each dimension analysis unit has a corresponding dimension accuracy. The Q dimension analysis units include a first pressure analysis unit, a second pressure analysis unit, and a temperature analysis unit corresponding to P types of environmental information, and a unit time displacement analysis unit, a vibration frequency analysis unit, a vibration peak value analysis unit, and an acceleration analysis unit corresponding to R types of operational information. This achieves the technical effect of constructing a comprehensive and accurate control analysis model for loss of control during electrically controlled drilling, thereby improving the control reliability of safe loss of control.

[0069] Step S400: Input the detection data set into the Q comprehensive analysis units to obtain Q first analysis results; input the Q data from the environmental information set and the operation information set into the Q dimensional analysis units respectively to obtain Q second analysis results;

[0070] Step S500: Perform a weighted calculation on the Q first analysis results and the Q second analysis results to obtain a comprehensive abnormal probability value and a comprehensive normal probability value. When the comprehensive abnormal probability value is greater than a preset abnormal threshold, control the target dropping structure to perform a dropping operation.

[0071] Furthermore, step S500 of this application also includes:

[0072] Step S510: Count the number of times the anomaly judgment result is abnormal and normal within the Q first analysis results and Q second analysis results, and obtain the number of abnormalities and the number of normalities;

[0073] Step S520: Sum the accuracy rates of the comprehensive analysis unit and dimensional analysis unit whose anomaly judgment results are abnormal in the first analysis result and the second analysis result to obtain the first summed accuracy rate; sum the accuracy rates of the comprehensive analysis unit and dimensional analysis unit whose anomaly judgment results are normal in the first analysis result and the second analysis result to obtain the second summed accuracy rate.

[0074] Step S530: The first and second sum accuracy rates are weighted and calculated with the abnormal and normal times respectively to obtain the comprehensive abnormal and normal probability values.

[0075] Specifically, Q data are extracted from the environmental information set and the operation information set. Then, the detection data set is input into Q comprehensive analysis units of the trip control analysis model to obtain Q first analysis results. The Q data are input into corresponding Q dimensional analysis units of the trip control analysis model as input information to obtain Q second analysis results.

[0076] The Q data include a first pressure average, a second pressure average, a temperature average in the environmental information set, and a unit time displacement average, a vibration frequency average, a vibration peak average, and an acceleration average in the operation information set. Each first analysis result includes whether the drilling operation of the target drilling equipment is abnormal and whether the trip operation needs to be performed on the target trip structure. Each second analysis result includes whether the drilling operation of the target drilling equipment is abnormal and whether the trip operation needs to be performed on the target trip structure.

[0077] Further, when the first analysis result is that the drilling operation of the target drilling equipment is abnormal and the trip operation needs to be performed on the target trip structure, an abnormal mark is added to the first analysis result. Otherwise, a normal mark is added to the first analysis result. Similarly, when the second analysis result is that the drilling operation of the target drilling equipment is abnormal and the trip operation needs to be performed on the target trip structure, an abnormal mark is added to the second analysis result. Otherwise, a normal mark is added to the second analysis result. Based on this, the Q first analysis results and the Q second analysis results are traversed to add abnormal marks / normal marks. The number of abnormal marks and normal marks based on the Q first analysis results and the Q second analysis results is counted to obtain the abnormal and normal times. The abnormal time includes the sum of the number of abnormal marks corresponding to the Q first analysis results and the Q second analysis results. The normal time includes the sum of the number of normal marks corresponding to the Q first analysis results and the Q second analysis results.

[0078] Further, the accuracy rates of the comprehensive analysis unit of the first analysis result corresponding to the abnormal label and the accuracy rates of the dimension analysis unit of the second analysis result corresponding to the abnormal label are added to obtain a first added accuracy rate. Similarly, the accuracy rates of the comprehensive analysis unit of the first analysis result corresponding to the normal label and the accuracy rates of the dimension analysis unit of the second analysis result corresponding to the normal label are added to obtain a second added accuracy rate. The first added accuracy rate includes the sum of the comprehensive accuracy rate of the comprehensive analysis unit of the first analysis result corresponding to the abnormal label and the dimension accuracy rate of the dimension analysis unit of the second analysis result corresponding to the abnormal label. The second added accuracy rate includes the sum of the comprehensive accuracy rate of the comprehensive analysis unit of the first analysis result corresponding to the normal label and the dimension accuracy rate of the dimension analysis unit of the second analysis result corresponding to the normal label.

[0079] Further, the first added accuracy rate is multiplied by the number of abnormalities to obtain an abnormal weighted accuracy rate. The second added accuracy rate is multiplied by the number of normal times to obtain a normal weighted accuracy rate. The abnormal weighted accuracy rate and the normal weighted accuracy rate are added to obtain a weighted calculation accuracy rate. Then, the ratio between the abnormal weighted accuracy rate and the weighted calculation accuracy rate is output as a comprehensive abnormal probability value. The ratio between the normal weighted accuracy rate and the weighted calculation accuracy rate is output as a comprehensive normal probability value. Then, it is judged whether the comprehensive abnormal probability value is greater than a preset abnormal threshold value. If the comprehensive abnormal probability value is greater than the preset abnormal threshold value, the control target releases the structure to perform a release operation. The preset abnormal threshold value includes a comprehensive abnormal probability threshold value determined by the electric control drilling safety release control system, which can be set by a person skilled in the art, for example, 60%. The reliable comprehensive abnormal probability value is determined by the release control analysis model, thereby improving the control accuracy of the electric control drilling safety release and avoiding the technical effect of the misoperation of the electric control drilling safety release.

[0080] In summary, the electric control drilling safety release control method provided by the present application has the following technical effects:

[0081] 1. In the process of drilling by using a target drilling equipment, a set of environmental information, a set of operation information and a set of detection data are obtained by Q detection devices arranged on the target drilling equipment; a releasing control analysis model is constructed based on drilling detection data in a historical time; the set of detection data is input into Q comprehensive analysis units of the releasing control analysis model to obtain Q first analysis results, and the set of environmental information and the set of operation information are input into Q dimension analysis units of the releasing control analysis model to obtain Q second analysis results; the Q first analysis results and the Q second analysis results are weighted calculated to obtain a comprehensive abnormal probability value and a comprehensive normal probability value, and when the comprehensive abnormal probability value is greater than a preset abnormal threshold, a target releasing structure is controlled to perform a releasing operation. The technical effects of improving the control accuracy of the electrically controlled drilling safety releasing, improving the disengagement success rate of the electrically controlled drilling safety releasing, avoiding the misoperation of the electrically controlled drilling safety releasing and improving the control quality of the electrically controlled drilling safety releasing are achieved.

[0082] 2. The Q comprehensive analysis units with high accuracy and generalization performance are constructed by training a plurality of sample detection data sets and sample abnormal judgment result sets, so as to improve the accuracy of the releasing control analysis model.

[0083] Embodiment Two

[0084] Based on the same inventive concept as the above-mentioned embodiment, the present application also provides an electrically controlled drilling safety releasing control system, please refer to the accompanying drawings Figure 3 , the system comprises:

[0085] A drilling detection module 11 is configured to detect P kinds of environmental information in a well and R kinds of operation information in a target drilling equipment by Q detection devices arranged on the target drilling equipment in the process of drilling by using the target drilling equipment, Q, P and R are integers greater than 1, Q is the sum of P and R, the P kinds of environmental information include at least one of a first pressure, a second pressure and a temperature, the R kinds of operation information include at least one of a unit time displacement, a vibration frequency, a vibration peak value and an acceleration, and the target drilling equipment includes a target releasing structure.

[0086] A detection data set obtaining module 12 is configured to collect a set of environmental information and a set of operation information and obtain a set of detection data.

[0087] A construction module 13 is configured to construct a releasing control analysis model based on drilling detection data in a historical time, wherein the releasing control analysis model is constructed based on ensemble learning and includes Q comprehensive analysis units and Q dimension analysis units.

[0088] The comprehensive analysis module 14 is configured to input the detection data set into the Q comprehensive analysis units, obtain Q first analysis results, input Q data in the environment information set and the operation information set into the Q dimension analysis units respectively, and obtain Q second analysis results;

[0089] The hand-off control module 15 is configured to perform weighted calculation on the Q first analysis results and the Q second analysis results, obtain a comprehensive abnormal probability value and a comprehensive normal probability value, and control the target hand-off structure to perform a hand-off operation when the comprehensive abnormal probability value is greater than a preset abnormal threshold.

[0090] Further, the system further comprises:

[0091] The first execution module is configured to detect the P kinds of environment information and the R kinds of operation information through the Q detection devices in a plurality of continuous time windows at a preset frequency, and obtain a plurality of initial environment information sets and a plurality of initial operation information sets;

[0092] The mean value calculation module is configured to perform mean value calculation on data of the same kind of environment information and data of the same kind of operation information in the plurality of initial environment information sets and the plurality of initial operation information sets respectively, and obtain the environment information set and the operation information set;

[0093] The second execution module is configured to integrate the environment information set and the operation information set, and obtain the detection data set.

[0094] Further, the system further comprises:

[0095] The sample data obtaining module is configured to obtain a plurality of sample detection data sets, a plurality of sample environment information sets, a plurality of sample operation information sets, and a sample abnormality judgment result set based on drilling detection data in a historical time, and the sample abnormality judgment result includes a judgment result of whether drilling operation needs to be performed.

[0096] The comprehensive analysis unit construction module is configured to use the plurality of sample detection data sets and the sample abnormality judgment result set as construction data to construct the Q comprehensive analysis units;

[0097] The dimension analysis unit construction module is configured to use the plurality of sample environment information sets, the plurality of sample operation information sets, and the sample abnormality judgment result set as construction data to construct the Q dimension analysis units;

[0098] A third execution module is configured to integrate the Q comprehensive analysis units and the Q dimension analysis units to obtain the hand control analysis model.

[0099] Further, the system further comprises:

[0100] A division and combination module is configured to divide and combine data in the plurality of sample detection data sets and sample anomaly judgment result sets to obtain K sets of construction data.

[0101] A first weight distribution module is configured to distribute K first comprehensive weights to the K sets of construction data, and each first comprehensive weight is 1 / K.

[0102] A fourth execution module is configured to use the K sets of construction data to construct a first comprehensive analysis unit in the Q comprehensive analysis units based on a BP neural network.

[0103] A second comprehensive weight calculation module is configured to use the K sets of construction data to test a first comprehensive accuracy of the first comprehensive analysis unit, combine the K first comprehensive weights, and calculate K second comprehensive weights.

[0104] A fifth execution module is configured to use the K sets of construction data to construct a second comprehensive analysis unit in the Q comprehensive analysis units based on a BP neural network according to the K second comprehensive weights, wherein the training computing resource of each set of construction data is positively correlated with the size of the second comprehensive weight.

[0105] A sixth execution module is configured to continue to construct the Q comprehensive analysis units.

[0106] Further, the system further comprises:

[0107] A first dimension analysis unit construction module is configured to use sample environment information of a first environment information in the P kinds of environment information in the K sets of construction data and K sample anomaly judgment results as K sets of first dimension construction data, construct a first dimension analysis unit in the Q dimension analysis units based on a BP neural network, and train the first dimension analysis unit.

[0108] A first dimension accuracy obtaining module is configured to use the K sets of first dimension construction data to test the first dimension analysis unit to obtain a first dimension accuracy.

[0109] The first dimension weight determination module is configured to calculate K first dimension weights according to the first dimension accuracy and the K first comprehensive weights.

[0110] The second dimension analysis unit construction module is configured to use sample environment information of second environment information in the P kinds of environment information in the K sets of construction data and K sample anomaly judgment results as K sets of second dimension construction data, and construct a second dimension analysis unit in the Q dimension analysis units based on a BP neural network according to the K first dimension weights, wherein the training computing resource of each set of second dimension construction data is positively correlated with the size of the first dimension weight.

[0111] The seventh execution module is configured to test the second dimension analysis unit by using the K sets of second dimension construction data, obtain a second dimension accuracy, calculate K second dimension weights, and continue to construct the Q dimension analysis units.

[0112] The K sets of construction data are used to test the first comprehensive accuracy of the first comprehensive analysis unit, the K second comprehensive weights are calculated in combination with the K first comprehensive weights, and the following formula is used:

[0113]

[0114] W2(B1) is the second comprehensive weight of the first set of construction data B1, W1(B1) is the first comprehensive weight of the first set of construction data, Z1 is the first accuracy of the first comprehensive analysis unit, T is the number of the comprehensive analysis units that are currently constructed, and T=1, 2, 3…Q.

[0115] Further, the system further comprises:

[0116] The statistical module is configured to count the number of times of abnormal and normal anomaly judgment results in the Q first analysis results and the Q second analysis results, and obtain abnormal times and normal times.

[0117] The accuracy rate adding module is configured to add the accuracy rates of the comprehensive analysis units and the dimension analysis units in which the anomaly judgment results are abnormal in the output first analysis results and the second analysis results, and obtain a first added accuracy rate, and add the accuracy rates of the comprehensive analysis units and the dimension analysis units in which the anomaly judgment results are normal in the output first analysis results and the second analysis results, and obtain a second added accuracy rate.

[0118] The weighting calculation module is configured to perform weighting calculation on the first and second sum accuracies and the abnormal and normal times respectively to obtain the comprehensive abnormal probability value and the comprehensive normal probability value.

[0119] The electrically-controlled well drilling safety release control system provided by the embodiment of the present application can perform the electrically-controlled well drilling safety release control method provided by any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the performing method.

[0120] The various modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be implemented; in addition, the specific names of the various functional modules are only for the convenience of mutual differentiation, and are not used to limit the protection scope of the present application.

[0121] The present application provides an electrically-controlled well drilling safety release control method, wherein the method is applied to an electrically-controlled well drilling safety release control system, and the method comprises the following steps: in the process of drilling by using a target well drilling device, obtaining an environment information set, an operation information set and a detection data set by using Q detection devices arranged on the target well drilling device; constructing a release control analysis model based on drilling detection data in a historical time; inputting the detection data set into Q comprehensive analysis units of the release control analysis model to obtain Q first analysis results, and inputting the environment information set and the operation information set into Q dimension analysis units of the release control analysis model to obtain Q second analysis results; performing weighting calculation on the Q first analysis results and the Q second analysis results to obtain a comprehensive abnormal probability value and a comprehensive normal probability value; and when the comprehensive abnormal probability value is greater than a preset abnormal threshold value, controlling the target release structure to perform a release operation. The technical problems of poor control accuracy of the electrically-controlled well drilling safety release, low release success rate of the electrically-controlled well drilling safety release and high possibility of misoperation of the electrically-controlled well drilling safety release, which result in poor control effect of the electrically-controlled well drilling safety release, are solved. The technical effects of improving the control accuracy of the electrically-controlled well drilling safety release, improving the release success rate of the electrically-controlled well drilling safety release, avoiding misoperation of the electrically-controlled well drilling safety release and improving the control quality of the electrically-controlled well drilling safety release are achieved.

[0122] Note that the above are only the preferred embodiments of the present application and the technical principles applied. Those skilled in the art will understand that the present application is not limited to the specific embodiments described herein, and those skilled in the art can make various obvious changes, readjustments and substitutions without departing from the protection scope of the present application. Therefore, although the present application has been described in more detail through the above embodiments, the present application is not limited to the above embodiments, and can include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. A method for safe release control in electrically controlled drilling, characterized in that, The method includes: During the drilling process using the target drilling equipment, Q detection devices installed on the target drilling equipment detect P types of environmental information in the well and R types of operational information in the target drilling equipment, where Q, P, and R are integers, and Q is the sum of P and R. The P types of environmental information include at least one of first pressure, second pressure, and temperature. The R types of operational information include at least one of displacement per unit time, vibration frequency, vibration peak value, and acceleration. The target drilling equipment includes a target release structure. Collect and obtain a set of environmental information and a set of operational information, and obtain a set of detection data; Based on drilling detection data over a historical period, a loss control analysis model is constructed. The loss control analysis model is constructed based on ensemble learning and includes Q comprehensive analysis units and Q dimensional analysis units. The detection data set is input into the Q comprehensive analysis units to obtain Q first analysis results. The Q data points in the environmental information set and the operation information set are respectively input into the Q dimensional analysis units to obtain Q second analysis results. The Q first analysis results and the Q second analysis results are weighted and calculated to obtain a comprehensive anomaly probability value and a comprehensive normal probability value. When the comprehensive anomaly probability value is greater than a preset anomaly threshold, the target dropping structure is controlled to perform a dropping operation. Among them, a loss control analysis model is constructed based on drilling detection data over a historical period, including: Based on drilling detection data over a historical period, multiple sets of sample detection data, multiple sets of sample environmental information, multiple sets of sample operational information, and a set of sample anomaly judgment results are obtained. The sample anomaly judgment results include the judgment results of whether an anomaly has occurred in the drilling operation and whether the operation needs to be abandoned. The Q comprehensive analysis units are constructed using the multiple sample detection data sets and the sample anomaly judgment result sets as construction data. The Q-dimensional analysis units are constructed using the set of multiple sample environment information, the set of multiple sample operation information, and the set of sample anomaly judgment results as construction data. By integrating the Q comprehensive analysis units and the Q dimensional analysis units, the drop-out control analysis model is obtained; Specifically, the multiple sample detection data sets and the anomaly judgment result set are used as construction data to construct the Q comprehensive analysis units, including: The data within the multiple sample detection data sets and sample anomaly judgment result sets are divided and combined to obtain K sets of constructed data; K first comprehensive weights are assigned to the K groups of constructed data, and each first comprehensive weight is 1 / K; Using the K sets of constructed data, and based on a BP neural network, the first comprehensive analysis unit among the Q comprehensive analysis units is constructed and trained. Using the K sets of constructed data, the first comprehensive accuracy of the first comprehensive analysis unit is tested and obtained. Combined with the K first comprehensive weights, K second comprehensive weights are calculated. Using the K sets of constructed data, based on the BP neural network, and according to the K second comprehensive weights, the second comprehensive analysis units within the Q comprehensive analysis units are constructed and trained, wherein the training computing power resources of each set of constructed data are positively correlated with the magnitude of the second comprehensive weights; Continue constructing to obtain the Q comprehensive analysis units; Specifically, the Q first analysis results and the Q second analysis results are weighted and calculated to obtain a comprehensive anomaly probability value and a comprehensive normal probability value, including: Count the number of times the anomaly judgment result is abnormal and normal within the Q first analysis results and Q second analysis results, and obtain the number of abnormalities and the number of normalities; The accuracy rates of the comprehensive analysis unit and the dimensional analysis unit whose anomaly judgment results are abnormal within the first analysis result and the second analysis result are summed to obtain the first summed accuracy rate. The accuracy rates of the comprehensive analysis unit and the dimensional analysis unit whose anomaly judgment results are normal within the first analysis result and the second analysis result are summed to obtain the second summed accuracy rate. The first summation accuracy and the second summation accuracy are used to perform weighted calculations with the number of abnormal occurrences and the number of normal occurrences, respectively, to obtain the comprehensive abnormal probability value and the comprehensive normal probability value.

2. The method according to claim 1, characterized in that, Collect and obtain a set of environmental information and a set of operational information, and obtain a set of detection data, including: According to a preset frequency, in multiple consecutive time windows, the P types of environmental information and R types of operational information are detected by the Q detection devices to obtain multiple sets of initial environmental information and multiple sets of initial operational information. The average values ​​of the data of the same type of environmental information and the data of the same type of operational information within the multiple initial environmental information sets and multiple initial operational information sets are calculated respectively to obtain the environmental information set and the operational information set; The environmental information set and the operational information set are integrated to obtain the detection data set.

3. The method according to claim 1, characterized in that, Using the multiple sets of sample environment information, multiple sets of sample operation information, and sets of sample anomaly judgment results as construction data, the Q-dimensional analysis units are constructed, including: Using the sample environmental information of the first environmental information among the P types of environmental information in the K groups of constructed data and the anomaly judgment results of K samples as the first dimension construction data of the K groups, the first dimension analysis unit in the Q dimension analysis units is constructed and trained based on the BP neural network. The data is constructed using the K groups of the first dimension, and the first dimension analysis unit is tested to obtain the accuracy of the first dimension. Based on the accuracy of the first dimension and the K first comprehensive weights, the K first dimension weights are calculated. The sample environmental information of the second environmental information in the P types of environmental information within the K groups of constructed data and the K sample anomaly judgment results are used as the K groups of second dimension constructed data. Based on the BP neural network, according to the K first dimension weights, the second dimension analysis units within the Q dimension analysis units are constructed and trained. The training computing power resources of each group of second dimension constructed data are positively correlated with the magnitude of the first dimension weights. The second dimension analysis unit is tested using the K sets of second dimension construction data to obtain the second dimension accuracy, and K second dimension weights are calculated. The Q dimension analysis units are then constructed.

4. The method according to claim 1, characterized in that, Using the K sets of constructed data, the first comprehensive accuracy of the first comprehensive analysis unit is tested and obtained. Combined with the K first comprehensive weights, K second comprehensive weights are calculated as follows: ； in, Build data for the first group The second comprehensive weight, The first comprehensive weight is used to construct the first set of data. Let T be the first accuracy of the first comprehensive analysis unit, and T be the number of comprehensive analysis units that have been constructed so far, T=1,2,3…Q.

5. A safe release control system for electrically controlled drilling, characterized in that, The system is used to perform the method according to any one of claims 1 to 4, the system comprising: A drilling detection module is used to detect P types of environmental information and R types of operational information in the well during drilling using a target drilling rig, by using Q detection devices installed on the target drilling rig. Q, P, and R are integers, and Q is the sum of P and R. The P types of environmental information include at least one of a first pressure, a second pressure, and temperature. The R types of operational information include at least one of displacement per unit time, vibration frequency, vibration peak value, and acceleration. The target drilling rig includes a target release structure. A detection data set acquisition module is used to collect and acquire environmental information sets and operational information sets, and to obtain a detection data set; The construction module is used to build a loss control analysis model based on drilling detection data within a historical time period. The loss control analysis model is built based on ensemble learning and includes Q comprehensive analysis units and Q dimensional analysis units. The comprehensive analysis module is used to input the detection data set into the Q comprehensive analysis units to obtain Q first analysis results, and to input Q data from the environmental information set and the operation information set into the Q dimensional analysis units to obtain Q second analysis results. The drop control module is used to perform weighted calculations on the Q first analysis results and the Q second analysis results to obtain a comprehensive abnormal probability value and a comprehensive normal probability value. When the comprehensive abnormal probability value is greater than a preset abnormal threshold, the module controls the target drop structure to perform a drop operation.

Images