Abnormality detection system and abnormality detection method

Abstract

An anomaly detection system is a system that detects an anomaly, the anomaly detection system including an acquisition unit for preliminary preparation for acquiring measurement values for preliminary preparation of first and second parameters, an estimation model generation unit for generating an estimation model and a criterion used to detect an anomaly based on the measurement values for preliminary preparation, an acquisition unit for anomaly detection for acquiring measurement values for anomaly detection of the first and second parameters, an estimation value calculation unit for calculating an estimation value of the second parameter from the measurement value for anomaly detection of the first parameter by using the generated estimation model, and an anomaly detection unit for comparing the measurement value for anomaly detection of second parameter with the estimation value of the parameter, and detecting an anomaly based on a comparison result and the generated criteria.

Description

TECHNICAL FIELD

[0001] The present invention relates to an anomaly detection system and an anomaly detection method.BACKGROUND ART

[0002] In the related art, wells have been drilled for various purposes. Patent Literature 1 discloses that parameters related to drilling are monitored to detect an anomaly.CITATION LISTPatent Literature[Patent Literature 1] European Patent No. 2404031SUMMARY OF INVENTIONTechnical Problem

[0004] However, as disclosed in Patent Literature 1, simply using parameters related to drilling may not necessarily enable accurate detection of an anomaly, such as detection of sticking during drilling. Similar problems may also arise in the detection of an anomaly in situations other than well drilling.

[0005] An embodiment of the present invention has been made in view of the above, and an object thereof is to provide an anomaly detection system and an anomaly detection method which are capable of detecting an anomaly in drilling or the like with high accuracy.Solution to Problem

[0006] In order to achieve the above-mentioned object, an anomaly detection system according to one embodiment of the present invention is an anomaly detection system that detects an anomaly, the anomaly detection system including an acquisition means for preliminary preparation for acquiring measurement values for preliminary preparation of first and second parameters in a target different from an anomaly detection target, an estimation model generation means for generating, based on the measurement values acquired by the acquisition means for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly, an acquisition means for anomaly detection for acquiring measurement values for anomaly detection of the first and second parameters in an anomaly detection target, an estimation value calculation means for calculating an estimation value of the second parameter from the measurement value of the first parameter acquired by the acquisition means for anomaly detection, by using the estimation model generated by the estimation model generation means, and an anomaly detection means for comparing the measurement value of the second parameter acquired by the acquisition means for anomaly detection with the estimation value of the second parameter calculated by the estimation value calculation means, and detecting an anomaly in the target based on a comparison result and the criterion generated by the estimation model generation means.

[0007] In the anomaly detection system of one embodiment of the present invention, the measurement value of the second parameter is compared with the estimation value of the second parameter, and an anomaly is detected based on the comparison result and the criterion. In addition, when the estimation model used to detect an anomaly is generated from the measurement values for preliminary preparation, the criterion used to detect an anomaly is also generated. Thereby, an anomaly can be detected in accordance with an appropriate criterion. As a result, according to the anomaly detection system of one embodiment of the present invention, it is possible to detect an anomaly in drilling or the like with high accuracy based on comparison between the measurement value and the estimation value of the second parameter.

[0008] In the anomaly detection system, the anomaly detection target may be drilling of a well. According to this configuration, it is possible to detect an anomaly in drilling with high accuracy.

[0009] The estimation model generation means may calculate the estimation value of the second parameter from the measurement value for preliminary preparation of the first parameter by using the generated estimation model, calculate a comparison value based on the comparison between the measurement value for preliminary preparation of the second parameter and the calculated estimation value of the second parameter, and generate the criterion from a distribution of the comparison values. According to this configuration, a criterion can be generated appropriately and reliably, and as a result, an anomaly can be detected appropriately and reliably with high accuracy.

[0010] The first and second parameters may be parameters that are measurable outside the well. According to this configuration, it is possible to easily acquire measurement values of parameters and to easily detect an anomaly with high accuracy.

[0011] The acquisition means for anomaly detection may acquire measurement values for anomaly detection of the first and second parameters at a plurality of timings, the estimation value calculation means may calculate estimation values of the second parameter at the plurality of timings, and the anomaly detection means may compare the measurement values and the estimation values of the second parameter at the plurality of timings, and detect an anomaly in drilling based on a plurality of comparison results. According to this configuration, it is possible to detect an anomaly with higher accuracy.

[0012] The first parameter may be weight on bit of a drill bit used for drilling, the second parameter may be rotational torque of a drill pipe used for drilling, and the estimation model may be a linear regression model. The first parameter may be a parameter related to an injection flow rate of a drilling fluid used for drilling and a length of a drill pipe used for drilling, the second parameter may be an injection pressure of the drilling fluid used for drilling, and the estimation model may be a nonlinear regression model. According to these configurations, an anomaly can be detected appropriately and reliably with high accuracy.

[0013] The acquisition means for preliminary preparation may acquire measurement values for preliminary preparation of the first and second parameters according to a stage of drilling, the estimation model generation means may generate an estimation model and a criterion according to the stage of drilling, the acquisition means for anomaly detection may acquire measurement values for anomaly detection of the first and second parameters according to the stage of drilling, the estimation value calculation means may calculate the estimation value of the second parameter by using the estimation model according to the stage of drilling, and the anomaly detection means may detect an anomaly according to the stage of drilling. According to this configuration, it is possible to detect an anomaly with higher accuracy.

[0014] Incidentally, one embodiment of the present invention can be described as an invention of an anomaly detection system as described above, and also as an invention of an anomaly detection method as described below. These are substantially identical inventions with only different categories, and they have similar actions and effects.

[0015] That is, an anomaly detection method according to one embodiment of the present invention is an anomaly detection method which is a method of operating an anomaly detection system that detects an anomaly, the anomaly detection method including an acquisition step for preliminary preparation of acquiring measurement values for preliminary preparation of first and second parameters in a target different from an anomaly detection target, an estimation model generation step of generating, based on the measurement values acquired in the acquisition step for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly, an acquisition step for anomaly detection of acquiring measurement values for anomaly detection of the first and second parameters in an anomaly detection target, an estimation value calculation step of calculating an estimation value of the second parameter from the measurement value of the first parameter acquired in the acquisition step for anomaly detection by using the estimation model generated in the estimation model generation step, and an anomaly detection step of comparing the measurement value of the second parameter acquired in the acquisition step for anomaly detection with the estimation value of the second parameter calculated in the estimation value calculation step, and detecting an anomaly in the target based on a comparison result and the criterion generated in the estimation model generation step. In the anomaly detection method, the anomaly detection target may be the drilling of a well.Advantageous Effects of Invention

[0016] According to an embodiment of the present invention, an anomaly in drilling or the like can be detected with high accuracy.BRIEF DESCRIPTION OF DRAWINGS

[0017] FIG. 1 is a diagram showing a functional configuration of an anomaly detection system according to an embodiment of the present invention.

[0018] FIG. 2 is a graph showing an example of anomaly detection by an anomaly detection system.

[0019] FIG. 3 is a flowchart showing an anomaly detection method which is a process executed by an anomaly detection system according to an embodiment of the present invention.DESCRIPTION OF EMBODIMENTS

[0020] Hereinafter, an embodiment of an anomaly detection system and an anomaly detection method according to the present invention will be described in detail with reference to the drawings. Note that in the description of the drawings, the same elements are denoted by the same reference numerals, and repeated descriptions will be omitted.

[0021] FIG. 1 shows a functional configuration of an anomaly detection system 10 according to the present embodiment. The anomaly detection system 10 is a system (device) that detects an anomaly in drilling of a well. Drilling is performed on, for example, the seabed surface or the ground. That is, drilling is, for example, offshore drilling or onshore drilling. The drilling itself can be the same as in the prior art. For example, a drill bit connected to a drill pipe rotates and drills by a force transmitted from the drill pipe. In addition, a drilling fluid is pumped into the well to be drilled to remove drilled cuttings from the well. The drilling fluid is, for example, mud water (mud) or seawater. The drilling may be performed in a vertical direction or in other directions (for example, in a horizontal direction). Note that the drilling does not necessarily have to be performed as described above, and may be performed by any method.

[0022] An anomaly to be detected by the anomaly detection system 10 is sticking (stuck). An anomaly to be detected by the anomaly detection system 10 may also be a predictive sign that sticking is about to occur. The anomaly detection system 10 may also detect an anomaly in the drilling of a well other than sticking and a prediction of sticking. The anomaly detection system 10 may also detect the risk of an anomaly. The anomaly detection system 10 acquires measurement values (actual values) (drilling data) of parameters in drilling and detects an anomaly based on the acquired measurement values, as described below.

[0023] The anomaly detection system 10 is specifically configured with a computer including hardware such as a central processing unit (CPU) and a memory. Each of the functions to be described below in the anomaly detection system 10 is exhibited through the operation of these components by programs or the like. Note that the anomaly detection system 10 may be implemented by a single computer or by a computer system configured by a plurality of computers connected to each other by a network.

[0024] Subsequently, the functions of the anomaly detection system 10 according to the present embodiment will be described. As shown in FIG. 1, the anomaly detection system 10 includes an acquisition unit 11 for preliminary preparation, an estimation model generation unit 12, an acquisition unit 13 for anomaly detection, an estimation value calculation unit 14, and an anomaly detection unit 15. The anomaly detection system 10 generates an estimation model and a criterion used to detect an anomaly in advance, and detects an anomaly using the generated estimation model and criterion. The acquisition unit 11 for preliminary preparation and the estimation model generation unit 12 are configured to generate the estimation model and criterion.

[0025] The acquisition unit 11 for preliminary preparation is an acquisition means for preliminary preparation for acquiring measurement values for preliminary preparation of first and second parameters in drilling different from an anomaly detection target. The first and second parameters may be parameters that can be measured outside the well.

[0026] The anomaly detection system 10 uses two types of parameters that can be measured in drilling, that is, a first parameter and a second parameter which are different from each other, to detect an anomaly. The estimation model is a model that estimates (predicts) a value of the second parameter from a value of the first parameter. That is, the estimation model is a model in which a value of the first parameter is an explanatory variable and a value of the second parameter is an objective variable. Thus, the second parameter can be estimated from the first parameter.

[0027] A combination of the first and second parameters is set in advance. Each of the first and second parameters (types) in the combination may be multiple. The first and second parameters may be parameters that can be measured outside the well to be drilled, such as on the ground or on a vessel outside the well (specifically, on a rig located on the ground or on the vessel). The parameters in the example below are those that can be measured outside the well. Note that any or all of the first and second parameters may be those that can only be measured inside the well (for example, parameters obtained by measuring information inside the well using sensors).

[0028] Specifically, the combination (first combination) is one in which the first parameter is weight on bit of the drill bit used for drilling, and the second parameter is rotational torque of the drill pipe used for drilling. The weight on bit of the drill bit is, for example, weight on bit (surface WOB) of the drill bit which is calculated from a hook load. The weight on bit of the drill bit may be weight on bit (downhole WOB) of the drill bit which is measured inside the well. Alternatively, the combination (second combination) is one in which the first parameter is a parameter related to an injection flow rate of the drilling fluid used for drilling and related to the length of the drill pipe used for drilling, and the second parameter is an injection pressure (head pressure) of the drilling fluid used for drilling. Here, the parameter related to the length of the drill pipe is, for example, the total length of the drill pipe or an equivalent thereof (for example, the depth of the drill bit). The length of the drill pipe (for example, the total length of the drill pipe) may be actually used as the parameter. However, since the length of the drill pipe is not often included in items measured by a drilling control device, the depth of the drill bit may be used as the parameter. Strictly speaking, the length of the drill pipe=the depth of the drill bit (=a length from a drill floor (location outside the well where a device performing drilling and the like are installed) to a lower end of the drill pipe)+a hook height (a length from the drill floor to an upper end of the drill pipe). However, since the hook height is not a large value, it may be neglected (that is, regarded as zero) in the calculation. In the following, a specific example of a parameter related to the length of the drill pipe is the depth of the drill bit.

[0029] The measurement values for preliminary preparation of the first and second parameters are measured, for example, by a measurement device installed in advance in a device for drilling, or the like. The acquisition unit 11 for preliminary preparation receives and acquires the measurements for preliminary preparation of the first and second parameters, for example, from the measurement device. Alternatively, the acquisition unit 11 for preliminary preparation may receive a user's input operation of the measurement values for preliminary preparation of the first and second parameters for the anomaly detection system 10, and acquire the measurement values. The acquisition unit 11 for preliminary preparation may also acquire measurement values for preliminary preparation of the first and second parameters by any method other than the above. The acquisition unit 11 for preliminary preparation acquires a sufficient number of measurement values for preliminary preparation of the first and second parameters for the generation of the estimation model by the estimation model generation unit 12.

[0030] The measurement values for preliminary preparation of the first and second parameters are measurement values when it is assumed that no anomalies in the drilling of the well have occurred. The measurement values for preliminary preparation may also be measurement values obtained at a timing prior to the timing at which an anomaly is to be detected in the well which is the anomaly detection target. For example, the measurement values of the first and second parameters that were not detected as having an anomaly by (the anomaly detection unit 15 of) the anomaly detection system 10 may be used as measurement values for preliminary preparation for the subsequent detection.

[0031] For example, when the above-mentioned first combination is used as the first and second parameters, measurement values obtained from 12.5 hours to 30 minutes prior to the timing at which an anomaly is to be detected in the well which is the anomaly detection target are used as measurement values for preliminary preparation. When the above-mentioned second combination of the first and second parameters is used, measurement values obtained from 25 hours to 1 hour prior to the timing at which an anomaly is to be detected in the well which is the anomaly detection target are used as measurement values for preliminary preparation. In addition, a time slot related to measurement values for preliminary preparation is not necessarily limited to the above-mentioned time slot. For example, the time slot may be a time slot ranging from several minutes to several days prior to the timing at which an anomaly is to be detected. In addition, the measurement values for preliminary preparation may also be measurement values at wells (for example, offset wells) other than the well which is an anomaly detection target.

[0032] In addition, the measurement values of the parameters used in the anomaly detection system 10 may be values obtained by correcting actually measured values by a conventional method. In addition, the measurement values of the parameters may be converted from other parameters. In these cases, correction or conversion may be performed in the functional units of the anomaly detection system 10. The acquisition unit 11 for preliminary preparation outputs the acquired measurement values for preliminary preparation of the first and second parameters to the estimation model generation unit 12.

[0033] The estimation model generation unit 12 is an estimation model generation means for generating, based on the measurement values acquired by the acquisition unit 11 for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly. The estimation model generation unit 12 may calculate an estimation value of the second parameter from the measurement value for preliminary preparation of the first parameter by using the generated estimation model, calculate a comparison value based on comparison between the measurement value for preliminary preparation of the second parameter and the calculated estimation value of the second parameter, and generate the criterion from a distribution of the comparison values.

[0034] The first parameter may be weight on bit of the drill bit used for the drilling, the second parameter may be rotational torque of the drill pipe used for the drilling, and the estimation model may be a linear regression model. The first parameter may be a parameter related to an injection flow rate of the drilling fluid used for the drilling and the length of the drill pipe used for the drilling, the second parameter may be an injection pressure of the drilling fluid used for the drilling, and the estimation model may be a nonlinear regression model.

[0035] The estimation model generation unit 12 inputs the measurement values for preliminary preparation of the first and second parameters from the acquisition unit 11 for preliminary preparation. The estimation model generation unit 12 generates an estimation model based on the input measurement values for preliminary preparation. The estimation model generation unit 12 stores a generation method for an estimation model in advance and generates an estimation model based on the stored generation method. The estimation model generated may reflect knowledge obtained in advance (for example, physical knowledge (equations of motion, or the like), experimental knowledge (experimental measurement values, or the like), and operational knowledge (other knowledge obtained in actual operations or experience)). In addition, the estimation model generated is a data-driven model that is generated from only available data as in the present embodiment.

[0036] For example, when the above-mentioned first combination is used as the first and second parameters, the estimation model generation unit 12 generates an estimation model that is a linear regression model by linear regression. The linear regression model generated in this case is, for example, the following formula(Rotational⁢ torque⁢ of⁢ drill⁢ pipe)=w⁢1+w⁢2⁢ (Weight⁢ on⁢ bit⁢ of⁢ drill⁢ bit)

[0037] The estimation model generation unit 12 calculates the parameters w1 and w2 of the estimation model by the least-squares method, for example, based on the measurement value for preliminary preparation of the weight on bit of the drill bit which is the first parameter, and the measurement value for preliminary preparation of the rotational torque of the drill pipe which is the second parameter. When the parameters w1 and w2 of the estimation model are calculated, the measurement value for preliminary preparation of the first parameter and the measurement value for preliminary preparation of the second parameter (for example, the measurement values at the same timing), which correspond to each other, are used as a set of measurement values (the same is true in other examples).

[0038] In the above formula, “w2 (weight on bit of drill bit)” represents a friction torque applied to the drill bit, and “w1” represents the sum of a friction torque with a pit wall which occurs in the drill pipe, a friction torque of a top drive system for rotating the drill pipe, and the like. In this manner, the above formula is based on physical knowledge.

[0039] The parameters w1 and w2 of the estimation model with physical meaning, which are calculated as described above, may be used to detect an anomaly (by a method other than that in the present embodiment). For example, an anomaly may be detected from changes in the parameters w1 and w2 at a plurality of timings.

[0040] In addition, when the above-mentioned second combination is used as the first and second parameters, the estimation model generation unit 12 generates an estimation model which is a nonlinear regression model by nonlinear regression. The linear regression model generated in this case is, for example, the following formula(Injection⁢ pressure⁢ of⁢ drilling⁢ fluid)=w⁢1⁢ (Injection⁢ flow⁢ rate⁢ of⁢ drilling⁢ fluid)w⁢2+w⁢3⁢ (Depth⁢ of⁢ drill⁢ bit)⁢ (Injection⁢ flow⁢ rate⁢ of⁢ drilling⁢ fluid)w⁢4+w⁢5

[0041] The estimation model generation unit 12 calculates the parameters w1 to w5 of the estimation model by Huber regression, for example, based on the measurement values for preliminary preparation of the injection flow rate of the drilling fluid and the depth of the drill bit, which are the first parameters, and the measurement value for preliminary preparation of the injection pressure of the drilling fluid, which is the second parameter.

[0042] Once the estimation model is generated, the estimation model generation unit 12 generates a criterion used to detect an anomaly by using the generated estimation model. For example, the estimation model generation unit 12 generates a threshold value used to detect an anomaly as the criterion. However, the criterion may be anything other than a threshold value, as long as it can be used to detect an anomaly. For example, the estimation model generation unit 12 generates a criterion as follows. The criterion generated may also reflect knowledge obtained in advance, similar to the estimation model.

[0043] The estimation model generation unit 12 calculates an estimation value of the second parameter from the measurement value for preliminary preparation of the first parameter by using the generated estimation model. The estimation model generation unit 12 compares the calculated estimation value of the second parameter with the measurement value for preliminary preparation of the second parameter to calculate a comparison value. In this case, the measurement value for preliminary preparation of the second parameter used for comparison corresponds to the measurement value for preliminary preparation of the first parameter used to calculate the estimation value of the second parameter (for example, they are measurement values at the same timing).

[0044] For example, the estimation model generation unit 12 calculates a difference value calculated by (the measurement value for preliminary preparation of the second parameter)−(the estimation value of the second parameter) as a comparison value. The estimation model generation unit 12 calculates a plurality of comparison values using a plurality of measurement values for preliminary preparation. The estimation model generation unit 12 generates a threshold value based on a distribution of the comparison values (errors). The measurement values for preliminary preparation used to generate a criterion may or may not be used to generate the estimation model.

[0045] For the injection pressure of the drilling fluid, the difference value, which is the above-mentioned comparison value, is close to a normal distribution. Thus, when the above-mentioned second combination is used as the first and second parameters, that is, when the second parameter is the injection pressure of the drilling fluid, the estimation model generation unit 12 may use a value based on a σ value (standard deviation), which indicates a variation of the above-mentioned comparison values, as a threshold value. For example, the threshold value may be a mean value of the comparison values+2σ (2σ value of an error distribution). The threshold value may also be a preset percentile value of the difference value.

[0046] For the rotational torque of the drill pipe, the difference value, which is the above-mentioned comparison value, becomes complicated, and does not exhibit a normal distribution. Thus, when the above-mentioned first combination is used as the first and second parameters, that is, when the second parameter is the rotational torque of the drill pipe, the estimation model generation unit 12 may use a value based on the percentile value of the difference value as a threshold value, instead of using the σ value. For example, the threshold value may be a preset percentile value of the difference value (for example, 99 percentile value).

[0047] The estimation model generation unit 12 may generate estimation models and criteria other than those described above. The estimation model generation unit 12 may also generate estimation models and criteria by methods other than those described above. For example, data assimilation or machine learning methods (for example, machine learning methods that take into account physical models shown in Physics Informed Neural Networks (Raissi et al., 2019)) may be used to generate the estimation model. Kernel density estimation may also be used to generate the criterion. The estimation model generation unit 12 outputs the generated estimation model to the estimation value calculation unit 14. The estimation model generation unit 12 outputs the generated criterion to the anomaly detection unit 15.

[0048] The acquisition unit 13 for anomaly detection, the estimation value calculation unit 14, and the anomaly detection unit 15 are configured to detect an anomaly using the estimation model and criterion generated as described above.

[0049] The acquisition unit 13 for anomaly detection is an acquisition means for anomaly detection for acquiring measurement values for anomaly detection of the first and second parameters in the drilling of an anomaly detection target. The acquisition unit 13 for anomaly detection may acquire measurement values for anomaly detection of the first and second parameters at a plurality of timings.

[0050] The measurement values for anomaly detection of the first and second parameters are measured by, for example, a measurement device installed in advance in a device for drilling in the same manner as the measurement values for preliminary preparation. The acquisition unit 13 for anomaly detection receives and acquires, for example, the measurement values for anomaly detection of the first and second parameters from the measurement device. Alternatively, the acquisition unit 13 for anomaly detection may receive a user's input operation of the measurement values for anomaly detection of the first and second parameters for the anomaly detection system 10, and acquire the measurement values. The acquisition unit 11 for preliminary preparation may also acquire the measurement values for anomaly detection of the first and second parameters by any method other than the above.

[0051] The drilling of an anomaly detection target may be, for example, drilling in a time slot with a time range (for example, a time range from 5 minutes to 1 hour). The acquisition unit 13 for anomaly detection may acquire the measurement values for anomaly detection of the first and second parameters at a plurality of timings (for example, timings at regular intervals in the time slot) included in the time slot with the time range. In this case, the anomaly detection system 10 detects an anomaly by using these plurality of measurement values for anomaly detection.

[0052] In order to make anomaly detection performed by the anomaly detection system 10 useful in avoiding hazards in drilling taking place, the acquisition of measurement values for anomaly detection by the acquisition unit 13 for anomaly detection may be performed in real time or close to it. The acquisition unit 13 for anomaly detection outputs the acquired measurement value for anomaly detection of the first parameter to the estimation value calculation unit 14. The acquisition unit 13 for anomaly detection outputs the acquired measurement value for anomaly detection of the second parameter to the anomaly detection unit 15.

[0053] The estimation value calculation unit 14 is an estimation value calculation means for calculating an estimation value of the second parameter from the measurement value of the first parameter acquired by the acquisition unit 13 for anomaly detection, using the estimation model generated by the estimation model generation unit 12. The estimation value calculation unit 14 may calculate estimation values of the second parameter at a plurality of timings.

[0054] The estimation value calculation unit 14 inputs and stores the estimation model from the estimation model generation unit 12. The estimation value calculation unit 14 inputs measurement values for anomaly detection of the first parameter from the acquisition unit 13 for anomaly detection. The estimation value calculation unit 14 calculates an estimation value of the second parameter from the input measurement value for anomaly detection of the first parameter using the stored estimation model. When a plurality of measurement values are input from the acquisition unit 13 for anomaly detection, the estimation value calculation unit 14 calculates an estimation value for each measurement value. The estimation value calculation unit 14 may calculate (predict) estimation values not for all measurement values, but for measurement values that are thinned out at regular intervals. The estimation value calculation unit 14 may also calculate an estimation value at time t from a measurement value at time t by using a static estimation model. Alternatively, the estimation value calculation unit 14 may calculate an estimation value at time t from measurement values that are time-series data (for example, measurement values from time t-a to time t) by using a dynamic estimation model. The estimation value calculation unit 14 outputs the calculated estimation value of the second parameter to the anomaly detection unit 15.

[0055] The anomaly detection unit 15 is an anomaly detection means for comparing the measurement value of the second parameter acquired by the acquisition unit 13 for anomaly detection with the estimation value of the second parameter calculated by the estimation value calculation unit 14, and detecting an anomaly in drilling based on the comparison results and the criterion generated by the estimation model generation unit 12. The anomaly detection unit 15 may compare the measurement values and the estimation values of the second parameter at a plurality of timings, and detect an anomaly in drilling based on a plurality of comparison results. For example, the anomaly detection unit 15 detects an anomaly in drilling as follows

[0056] The anomaly detection unit 15 inputs and stores the criterion from the estimation model generation unit 12. The anomaly detection unit 15 inputs the measurement value for anomaly detection of the second parameter from the acquisition unit 13 for anomaly detection. The anomaly detection unit 15 inputs the estimation value of the second parameter from the estimation value calculation unit 14. The anomaly detection unit 15 compares the input estimation value of the second parameter with the measurement value for anomaly detection of the second parameter to calculate a comparison value. In this case, the measurement value for anomaly detection of the second parameter used for comparison corresponds to the measurement value for anomaly detection of the first parameter used to calculate the estimation value of the second parameter in the estimation value calculation unit 14 (for example, it is the measurement value at the same timing). In addition, the comparison value calculated here is calculated in the same manner (for example, with the same formula) as the comparison value calculated by the estimation model generation unit 12.

[0057] For example, the anomaly detection unit 15 calculates a difference value calculated by (the measurement value for anomaly detection of the second parameter)−(the estimation value of the second parameter) as a comparison value. When there are a plurality of measurement values for anomaly detection of the second parameter and a plurality of estimation values of the second parameter, the anomaly detection unit 15 calculates a plurality of comparison values for each of the plurality of measurement values for anomaly detection and the plurality of estimation values.

[0058] The anomaly detection unit 15 detects an anomaly in drilling based on the calculated comparison values and stored criterion. For example, the anomaly detection unit 15 determines whether the calculated comparison value falls within a normal range indicated by the criterion. Specifically, the anomaly detection unit 15 compares the comparison value with the threshold value which is the criterion. For example, when the comparison value exceeds the threshold value, the anomaly detection unit 15 determines that there is an anomaly in the drilling. When the comparison value is equal to or less than the threshold value, the anomaly detection unit 15 determines that there is no anomaly in the drilling.

[0059] When there are a plurality of comparison values, the anomaly detection unit 15 may count the number of comparison values that exceed the threshold value and calculate the degree of anomaly, which is the degree to which an anomaly has occurred, based on the number of comparison values. For example, the anomaly detection unit 15 may calculate an anomaly degree a (t) in accordance with the following formula.a⁡(t)=1N⁢∑k=1NI[ε⁡(t-Δ⁢t⁡(k-1))>εth(t-Δ⁢t⁡(k-1))][Math. 1]Here,I[ε⁡(t)>εth(t)]={1,ε⁡(t)>εth(t)0,ε⁡(t)≤εth(t)

[0060] In the above formula, t is a timing (time), N is a total number of measurement values for anomaly detection, ε(t) is a difference value at a timing t, εth(t) is a threshold value, and Δt is a sampling period of measurement values for anomaly detection. The above anomaly degree a(t) is a ratio of the number of comparison values exceeding the threshold value to the total number of comparison values, and a larger value indicates a higher degree of anomaly.

[0061] The anomaly detection unit 15 may detect an anomaly in drilling by methods other than those described above, as long as it compares the measurement values and estimation values of the second parameter and detects an anomaly in drilling based on the comparison results and the above criterion. For example, the above-mentioned difference value may be evaluated using a statistical method or a method using machine learning other than the above-mentioned methods to detect an anomaly.

[0062] The anomaly detection unit 15 outputs information indicating the results of anomaly detection. For example, the anomaly detection unit 15 may output the information in a format (for example, display) that can be recognized by the user of the anomaly detection system 10. The user can cope with an anomaly in the drilling of the well when the anomaly is detected, with reference to the output from the anomaly detection system 10. The anomaly detection unit 15 may also output the information in other formats. For example, the anomaly detection unit 15 may transmit the information to another device.

[0063] FIG. 2 shows four example graphs of difference values for preliminary preparation (difference values for calculating threshold values), difference values for anomaly detection, and threshold values. In each of the graphs in FIG. 2, a horizontal axis represents the axis of a difference value and a threshold value, and a vertical axis represents the axis of a probability distribution (the degree to which a difference value appears). There is physical knowledge that an average increase or greater dispersion in the rotational torque of the drill pipe and the injection pressure of the drilling fluid is a predictive sign of sticking. This is knowledge under the condition that other drilling parameters are the same. This means that, when the first parameter is the same, the second parameter will increase on average. For example, when an injection flow rate is increased, the injection pressure will increase even when there is no anomaly. On the other hand, an increase in injection pressure while the injection flow rate remains the same is a predictive sign of sticking.

[0064] In the cases shown in the graphs in FIGS. 2(a) and 2(b), a certain number of difference values for anomaly detection exceed the threshold value, which is detected as an anomaly (specifically, a predictive sign of sticking) in the present embodiment. In the cases shown in the graphs in FIGS. 2(c) and 2(d), although a distribution of difference values for anomaly detection is different from a distribution of difference values for preliminary preparation, a certain number of difference values for anomaly detection does not exceed a threshold value and is not detected as an anomaly (specifically, a predictive sign of sticking) in the present embodiment. The cases shown in the graphs in FIGS. 2(c) and 2(d) do not usually indicate an anomaly and are appropriate for detection in the present embodiment. However, in general change point detection, all of (a) to (d) are taken as anomalies.

[0065] The values of the above parameters can vary depending on the stage of drilling. The stage of drilling is the operational stage of drilling, for example, the stage of drilling with the drill bit (drilling), the stage of lowering the drill bit down into the well (run-in-hole), and the stage of pulling the drill bit out of the well (pull-out-of-hole). In particular, during replacement of the drill bit, the values of the above parameters fluctuate significantly although such fluctuations have little physical significance. For this reason, when an attempt to detect an anomaly is made without distinguishing between the stages, there is a concern that the anomaly may not be detected appropriately. In consideration of the above, the functions of the functional units of the anomaly detection system 10 described above may be for each stage of drilling.

[0066] That is, the acquisition unit 11 for preliminary preparation may acquire measurement values for preliminary preparation of the first and second parameters according to the stage of drilling. The estimation model generation unit 12 may generate an estimation model and a criterion according to the stage of drilling. The acquisition unit 13 for anomaly detection may acquire measurement values for anomaly detection of the first and second parameters according to the stage of drilling. The estimation value calculation unit 14 may calculate an estimation value of the second parameter using an estimation model according to the stage of drilling. The anomaly detection unit 15 may detect an anomaly according to the stage of drilling.

[0067] For example, for each stage of drilling, information indicating the corresponding stage may be input to the anomaly detection system 10, and the functions of the functional units of the anomaly detection system 10 may be executed for the stage. The stage of drilling can be determined by the existing method such as determining the stage of drilling based on a pre-given conditional formula by using drilling data. The anomaly detection system 10 may determine the stage of drilling. In addition, when considering a static physical model as in the above-mentioned example, operations of excluding parts having a large dynamic impact (for example, operation parameters such as an injection flow rate) may be performed simultaneously. Alternatively, the functions of the functional units of the anomaly detection system 10 may be performed by excluding certain stages. These are the functions of the anomaly detection systems 10.

[0068] Subsequently, a flowchart of FIG. 3 is used to describe an anomaly detection method, which is a process (an operation method performed by the anomaly detection system 10) executed by the anomaly detection system 10 according to the present embodiment. This process is divided into two processes: a process before detecting an anomaly (S01, S02) and a process of detecting an anomaly (S03 to S07). The process before detecting an anomaly (S01, S02) may be performed before the process of detecting an anomaly (S03 to S07).

[0069] In the process before detecting an anomaly, the acquisition unit 11 for preliminary preparation acquires measurement values for preliminary preparation of the first and second parameters in drilling different from an anomaly detection target (S01, acquisition step for preliminary preparation). Next, the estimation model generation unit 12 generates, based on the measurement values for preliminary preparation of the first and second parameters, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly, (S02, estimation model generation step). The generated estimation model and criterion are used in the process of detecting an anomaly. The above is the process before detecting an anomaly.

[0070] In the process of detecting an anomaly, the acquisition unit 13 for anomaly detection acquires measurement values for anomaly detection of the first and second parameters in the drilling of an anomaly detection target (S03, acquisition step for anomaly detection). Next, the estimation value calculation unit 14 calculates an estimation value of the second parameter from the measurement value for anomaly detection of the first parameter by using an estimation model (S04, estimation value calculation step).

[0071] Next, the anomaly detection unit 15 compares the measurement value for anomaly detection of the second parameter and the estimation value of the second parameter (S05, anomaly detection step). Next, the anomaly detection unit 15 detects an anomaly in drilling based on a comparison result and the above-mentioned criterion (S06, anomaly detection step). Next, the result of anomaly detection is output from the anomaly detection unit 15 (S07). The above is a process executed by the anomaly detection system 10 according to the present embodiment.

[0072] In the present embodiment, a measurement value of a second parameter is compared with an estimation value of the second parameter, and an anomaly in drilling is detected based on a comparison result and a criterion. In addition, when an estimation model used to detect an anomaly is generated from measurement values for preliminary preparation, a criterion used to detect is also generated. Thereby, an anomaly can be detected in accordance with an appropriate criterion. As a result, according to the present embodiment, it is possible to detect an anomaly in drilling with high accuracy based on comparison between the measurement value and the estimation value of the second parameter. Further, in the present embodiment, it is possible to detect an anomaly without using data on past anomalies (labels of anomalies).

[0073] Further, as in the present embodiment, the estimation model generation unit 12 may calculate an estimation value of a second parameter from a measurement value for preliminary preparation of a first parameter by using an estimation model, calculate a comparison value based on comparison between the measurement values for preliminary preparation of the second parameter and the calculated estimation value of the second parameter, and generate a criterion from a distribution of the comparison values. According to this configuration, a criterion can be generated appropriately and reliably, and as a result, an anomaly can be detected appropriately and reliably with high accuracy. However, the generation of comparison does not necessarily have to be performed as described above, but may be performed based on measurement values for preliminary preparation of the first and second parameters.

[0074] In addition, as in the present embodiment, first and second parameters may be parameters that can be measured outside the well. According to this configuration, it is possible to easily acquire measurement values of parameters and to easily detect an anomaly with high accuracy. However, any or all of the parameters may not be parameters that can be measured outside the well.

[0075] In addition, as in the present embodiment, anomaly detection may be performed using measurement values for anomaly detection of first and second parameters at a plurality of timings. According to this configuration, it is possible to detect an anomaly with higher accuracy. However, measurement values for anomaly detection at a plurality of timings do not need to be used to detect an anomaly, and measurement values for anomaly detection at only one timing may be used.

[0076] In addition, as in the present embodiment, a first parameter may be weight on bit of a drill bit used for drilling, a second parameter may be rotational torque of a drill pipe used for drilling, and an estimation model may be a linear regression model. In addition, the first parameter may be a parameter related to an injection flow rate of a drilling fluid used for drilling and the length of a drill pipe used for drilling, the second parameter is an injection pressure of the drilling fluid used for drilling, and the estimation model may be a nonlinear regression model. According to these configurations, an anomaly can be detected appropriately and reliably with high accuracy. However, the first parameter, the second parameter, and the estimation model may be other than those mentioned above.

[0077] In addition, as in the present embodiment, an estimation model and a criterion may be generated in accordance with the stage of drilling, and they may be used to detect an anomaly according to the stage of drilling. According to this configuration, an anomaly can be detected with higher accuracy in consideration of the stage of drilling.

[0078] Note that, in the above-described embodiment, the anomaly detection system 10 detects an anomaly in the drilling of a well, but it may also detect an anomaly in any target other than the drilling of a well. That is, in the above-described embodiment, a target for detecting an anomaly is the drilling of a well, but a target for detecting an anomaly may be anything other than the drilling of a well. Also in this case, it is sufficient that anomaly detection is performed on a target in the same manner as in the above-described embodiment, with only an anomaly detection target and parameters according to the target being different from those in the above-described embodiment.

[0079] The anomaly detection system and the anomaly detection method according to the present disclosure have the following configurations.

[0080] [1] An anomaly detection system that detects an anomaly, the anomaly detection system including

[0081] an acquisition means for preliminary preparation for acquiring measurement values for preliminary preparation of first and second parameters in a target different from an anomaly detection target,

[0082] an estimation model generation means for generating, based on the measurement values acquired by the acquisition means for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly,

[0083] an acquisition means for anomaly detection for acquiring measurement values for anomaly detection of the first and second parameters in an anomaly detection target,

[0084] an estimation value calculation means for calculating an estimation value of the second parameter from the measurement value of the first parameter acquired by the acquisition means for anomaly detection, by using the estimation model generated by the estimation model generation means, and

[0085] an anomaly detection means for comparing the measurement value of the second parameter acquired by the acquisition means for anomaly detection with the estimation value of the second parameter calculated by the estimation value calculation means, and detecting an anomaly in the target based on a comparison result and the criterion generated by the estimation model generation means.

[0086] [2] The anomaly detection system according to [1], in which the anomaly detection target is drilling of a well.

[0087] [3] The anomaly detection system according to [2], in which the estimation model generation means calculates the estimation value of the second parameter from the measurement value for preliminary preparation of the first parameter by using the generated estimation model, calculates a comparison value based on the comparison between the measurement value for preliminary preparation of the second parameter and the calculated estimation value of the second parameter, and generates the criterion from a distribution of the comparison values.

[0088] [4] The anomaly detection system according to [2] or [3], in which the first and second parameters are parameters that are measurable outside the well.

[0089] [5] The anomaly detection system according to any one of [2] to [4], in which

[0090] the acquisition means for anomaly detection acquires measurement values for anomaly detection of the first and second parameters at a plurality of timings,

[0091] the estimation value calculation means calculates estimation values of the second parameter at the plurality of timings, and

[0092] the anomaly detection means compares the measurement values and the estimation values of the second parameter at the plurality of timings, and detects an anomaly in drilling based on the plurality of comparison results.

[0093] [6] The anomaly detection system according to any one of [2] to [5], in which

[0094] the first parameter is weight on bit of a drill bit used for drilling,

[0095] the second parameter is rotational torque of a drill pipe used for drilling, and

[0096] the estimation model is a linear regression model.

[0097] [7] The anomaly detection system according to any one of [2] to [6], in which

[0098] the first parameter is a parameter related to an injection flow rate of a drilling fluid used for drilling and a length of a drill pipe used for drilling,

[0099] the second parameter is an injection pressure of the drilling fluid used for drilling, and

[0100] the estimation model is a nonlinear regression model.

[0101] [8] The anomaly detection system according to any one of [2] to [7], in which

[0102] the acquisition means for preliminary preparation acquires measurement values for preliminary preparation of the first and second parameters according to a stage of drilling,

[0103] the estimation model generation means generates an estimation model and a criterion according to the stage of drilling,

[0104] the acquisition means for anomaly detection acquires measurement values for anomaly detection of the first and second parameters according to the stage of drilling,

[0105] the estimation value calculation means calculates the estimation value of the second parameter by using the estimation model according to the stage of drilling, and

[0106] the anomaly detection means detects an anomaly according to the stage of drilling.

[0107] [9] An anomaly detection method which is a method of operating an anomaly detection system that detects an anomaly, the anomaly detection method including

[0108] an acquisition step for preliminary preparation of acquiring measurement values for preliminary preparation of first and second parameters in a target different from an anomaly detection target;

[0109] an estimation model generation step of generating, based on the measurement values acquired in the acquisition step for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly;

[0110] an acquisition step for anomaly detection of acquiring measurement values for anomaly detection of the first and second parameters in an anomaly detection target;

[0111] an estimation value calculation step of calculating an estimation value of the second parameter from the measurement value of the first parameter acquired in the acquisition step for anomaly detection by using the estimation model generated in the estimation model generation step; and

[0112] an anomaly detection step of comparing the measurement value of the second parameter acquired in the acquisition step for anomaly detection with the estimation value of the second parameter calculated in the estimation value calculation step, and detecting an anomaly in the target based on a comparison result and the criterion generated in the estimation model generation step.

[0113] The anomaly detection method according to [9], in which the anomaly detection target is drilling of a well.REFERENCE SIGNS LIST10 Anomaly detection system, 11 Acquisition unit for preliminary preparation, 12 Estimation model generation unit, 13 Acquisition unit for anomaly detection, 14 Estimation value calculation unit, 15 Anomaly detection unit

Claims

1. An anomaly detection system that detects an anomaly, the anomaly detection system comprising circuitry configured to:acquire measurement values for preliminary preparation of first and second parameters in a target different from an anomaly detection target;generate, based on the acquired measurement values for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly;acquire measurement values for anomaly detection of the first and second parameters in an anomaly detection target;calculate an estimation value of the second parameter from the acquired measurement value for anomaly detection of the first parameter, by using the generated estimation model; andcompare the acquired measurement value for anomaly detection of the second parameter with the calculated estimation value of the second parameter, and detect an anomaly in the target based on a comparison result and the generated criterion.

2. The anomaly detection system according to claim 1,wherein the anomaly detection target is drilling of a well.

3. The anomaly detection system according to claim 2,wherein the circuitry calculates the estimation value of the second parameter from the measurement value for preliminary preparation of the first parameter by using the generated estimation model, calculates a comparison value based on the comparison between the measurement value for preliminary preparation of the second parameter and the calculated estimation value of the second parameter, and generates the criterion from a distribution of the comparison values.

4. The anomaly detection system according to claim 2,wherein the first and second parameters are parameters that are measurable outside the well.

5. The anomaly detection system according to claim 2,wherein the circuitry acquires measurement values for anomaly detection of the first and second parameters at a plurality of timings,calculates estimation values of the second parameter at the plurality of timings, andcompares the measurement values and the estimation values of the second parameter at the plurality of timings, and detects an anomaly in drilling based on a plurality of comparison results.

6. The anomaly detection system according to claim 2,wherein the first parameter is weight on bit of a drill bit used for drilling,the second parameter is rotational torque of a drill pipe used for drilling, andthe estimation model is a linear regression model.

7. The anomaly detection system according to claim 2,wherein the first parameter is a parameter related to an injection flow rate of a drilling fluid used for drilling and a length of a drill pipe used for drilling,the second parameter is an injection pressure of the drilling fluid used for drilling, andthe estimation model is a nonlinear regression model.

8. The anomaly detection system according to claim 2,wherein the circuitry acquires measurement values for preliminary preparation of the first and second parameters according to a stage of drilling,generates an estimation model and a criterion according to the stage of drilling,acquires measurement values for anomaly detection of the first and second parameters according to the stage of drilling,calculates the estimation value of the second parameter by using the estimation model according to the stage of drilling, anddetects an anomaly according to the stage of drilling.

9. An anomaly detection method which is a method of operating an anomaly detection system that detects an anomaly, the anomaly detection method comprising:acquiring measurement values for preliminary preparation of first and second parameters in a target different from an anomaly detection target;generating, based on the acquired measurement values for preliminary preparation, an estimation model for estimating a value of the second parameter from a value of the first parameter, and a criterion used to detect an anomaly;acquiring measurement values for anomaly detection of the first and second parameters in an anomaly detection target;calculating an estimation value of the second parameter from the acquired measurement value for anomaly detection of the first parameter by using the generated estimation model; andcomparing the acquired measurement value for anomaly detection of the second parameter with the calculated estimation value of the second parameter, and detecting an anomaly in the target based on a comparison result and the generated criterion.

10. The anomaly detection method according to claim 9,wherein the anomaly detection target is drilling of a well.

Images