A method and apparatus for diagnosing drilling equipment faults

Abstract

A method and apparatus for diagnosing drilling equipment faults, the method comprising: acquiring relevant monitoring data of the equipment in real time through sensors; wherein the monitoring data includes: vibration data and pressure data of the mud pump; determining that the mud pump has failed based on the pressure data of the mud pump in the monitoring data; performing slow feature analysis on the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals; and determining the diagnosis result of the mud pump fault based on the fast and slow feature signals.

Description

Technical Field

[0001] This article relates to the field of drilling technology, and in particular to a method and device for diagnosing drilling equipment faults. Background Technology

[0002] As a core piece of equipment in drilling operations, mud pumps are responsible for transporting drilling fluid, breaking rocks, removing cuttings, and cooling the drill bit. Their lifespan and reliability directly affect the safety and operating costs of drilling operations. However, mud pumps operate in harsh environments year-round and are easily affected by external factors. The mud pumps transport have a high sand content, high pressure, and strong corrosiveness, and the working environment is extremely poor. Under long-term operation, parts are prone to wear, seal failure, and pipeline blockage. In particular, the hydraulic end is prone to failure under long-term load, and even the shortest downtime can lead to huge economic losses.

[0003] Common failure modes of the hydraulic end of mud pumps include cylinder liner failure, piston failure, and intake / discharge valve malfunctions. The causes of these failures are varied, including valve seat damage, spring breakage, and leakage from the inlet / outlet valves. Mud pump failures are diverse, with vibration sources including inertial forces, pressure fluctuations, and vibrations from the plunger and cylinder liner. Considering the specific impact of each vibration source, mud pump structure type, and transmission characteristics on the vibration signal is difficult and impractical. Mud pump components are subject to wear and corrosion during operation, often manifesting as slow, imperceptible changes in the signal. However, vibration sources such as spring breakage often exhibit transient characteristics, making it difficult to effectively separate and characterize the gradual and transient changes.

[0004] Existing fault diagnosis methods for mud pumps mostly rely on single monitoring signals. While these methods can effectively assess and diagnose the overall operating status and fault modes of the pump, they cannot effectively determine the location and cause of the fault, making it difficult to pinpoint the faulty component and diagnose the problem. The monitoring parameters for different states of a mud pump vary, and weighted multi-source information fusion methods suffer from inconsistencies, making it difficult to effectively accommodate environmental characteristics from multiple parameters. Errors from a single sensor can significantly impact the diagnostic conclusions for the entire device, making these methods unsuitable for mud pumps operating under complex conditions, harsh environments, and with diverse fault modes.

[0005] Therefore, given the variable operating environment, there is an urgent need for an effective diagnostic method for mud pumps. Summary of the Invention

[0006] This application provides a method and apparatus for fault diagnosis of drilling equipment. The method includes: collecting relevant monitoring data of the equipment in real time through sensors; performing SFA analysis on the relevant monitoring data to obtain several signals containing fast and slow characteristics; and finally realizing equipment fault diagnosis under complex working conditions.

[0007] In a first aspect, this application provides a method for diagnosing faults in drilling equipment, the method comprising:

[0008] The equipment collects relevant monitoring data in real time through sensors; the monitoring data includes: mud pump vibration data and mud pump pressure data;

[0009] The mud pump malfunction was determined based on the mud pump pressure data in the monitoring data.

[0010] Slow feature analysis was performed on the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals;

[0011] The diagnosis result of the mud pump fault is determined based on the speed characteristic signal.

[0012] Secondly, embodiments of the present invention also provide a drilling equipment fault diagnosis device, characterized in that the device includes: a memory and a processor; the memory is used to store a program for performing drilling equipment fault diagnosis, and the processor is used to read and execute the program for performing drilling equipment fault diagnosis, and execute the method described in any one of the above embodiments.

[0013] Thirdly, embodiments of the present invention also provide a computer-readable storage medium storing a data processing program, wherein the data processing program is executed by a processor using the drilling equipment fault diagnosis method described in any of the above embodiments.

[0014] Compared with related technologies, this application provides a method and apparatus for fault diagnosis of drilling equipment. The method includes: acquiring relevant monitoring data of the equipment in real time through sensors; determining that the mud pump has failed based on the mud pump pressure data in the monitoring data; performing slow feature analysis on the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals; and determining the diagnostic result of the mud pump failure based on the fast and slow feature signals. This application acquires relevant monitoring data of the equipment in real time through sensors; performs slow feature analysis on the relevant monitoring data to obtain several signals containing fast and slow features, and finally realizes equipment fault diagnosis under complex working conditions.

[0015] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. Other advantages of this application can be realized and obtained by means of the solutions described in the description and the accompanying drawings. Attached Figure Description

[0016] The accompanying drawings are used to provide an understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0017] Figure 1 This is a flowchart of a drilling equipment fault diagnosis method according to an embodiment of this application;

[0018] Figure 2 This is a schematic diagram of a drilling equipment fault diagnosis device in some exemplary embodiments;

[0019] Figure 3 This is a schematic diagram of a drilling equipment fault diagnosis method in some exemplary embodiments;

[0020] Figure 4 This is a schematic diagram of a drilling equipment fault diagnosis device according to an embodiment of this application. Detailed Implementation

[0021] This application describes several embodiments, but these descriptions are exemplary and not restrictive, and it will be apparent to those skilled in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with, or may replace, any feature or element of any other embodiment.

[0022] This application includes and contemplates combinations of features and elements known to those skilled in the art. The embodiments, features, and elements disclosed in this application may also be combined with any conventional features or elements to form a unique inventive scheme as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive schemes to form another unique inventive scheme as defined by the claims. Therefore, it should be understood that any feature shown and / or discussed in this application may be implemented individually or in any suitable combination. Therefore, the embodiments are not limited except by the limitations imposed by the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

[0023] Furthermore, in describing representative embodiments, the specification may have presented methods and / or processes as a specific sequence of steps. However, the method or process should not be limited to the specific order of steps described herein, to the extent that it does not depend on such a specific order. As will be understood by those skilled in the art, other sequences of steps are also possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation of the claims. Moreover, the claims concerning the method and / or process should not be limited to the steps performed in the written order, and those skilled in the art will readily understand that these orders can be varied and still remain within the spirit and scope of the embodiments of this application.

[0024] This invention provides a method for diagnosing drilling equipment faults, such as... Figure 1 As shown, the method includes steps S100-S130:

[0025] S100: Collects relevant monitoring data of the equipment in real time through sensors;

[0026] S110: Based on the monitoring data, it is determined that the mud pump has malfunctioned;

[0027] S120: Perform slow feature analysis on the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals;

[0028] S130: Determine the diagnostic result of the mud pump fault based on the speed characteristic signal.

[0029] In this embodiment, the relevant monitoring data of the device can be collected in real time through sensors, as follows: Figure 2 The drilling equipment fault diagnosis system shown is equipped with multiple sensors for data acquisition. The sensor locations are listed in Table 1, which shows the key measuring points of the mud pump. The monitoring data that can be collected in real time by the sensors includes mud pump vibration data and mud pump pressure data.

[0030] Table 1

[0031] 1 mud pump hydraulic end 6 Add vibration sensor 2 mud pump discharge pipe 1 Add pressure sensor

[0032] like Figure 2 As shown, the drilling equipment fault diagnosis system can be divided into edge and cloud components. The edge component mainly performs tasks such as data acquisition, signal processing, fault detection, and data transmission, and displays the diagnostic conclusions from the cloud through a human-machine interface. The cloud component performs data storage, data management, data analysis, fault location, and fault diagnosis, realizing functions such as information integration, status monitoring, and alarm handling. The on-site drilling equipment fault diagnosis system is deployed in the cloud, and the backend of the fault diagnosis system and data analysis are completed by the cloud server.

[0033] Edge end: Sensors are installed at the measuring points as shown in Table 1. The sensor system module consists of vibration sensors and pressure sensors, which are connected to the data acquisition component to collect field data. The data acquisition component obtains the control parameters of the monitoring and control system, performs signal processing and fault detection, and collects the data to the cloud server for storage, management, and analysis. On-site personnel obtain information on normal equipment operation and accidents, as well as alarm handling, through the human-machine interface, which guides them to take relevant measures.

[0034] Cloud-based architecture: The cloud-based architecture mainly consists of a cloud database and cloud servers. The cloud database primarily stores, backs up, and manages edge data, while the cloud servers are mainly used for data analysis, fault location, fault diagnosis, and system deployment. The system's front-end is primarily built using Vue to display the status monitoring of mud pumps and the results of fault diagnosis. The back-end data analysis component performs data processing, SFA analysis (i.e., slow feature analysis), XGBoost feature analysis, SVM fault diagnosis, and DS evidence-based decision fusion, enabling the updating of models for mud pump fault detection, fault location, and fault diagnosis.

[0035] In one exemplary embodiment, the process of determining a mud pump malfunction based on monitoring data is as follows: Figure 3 As shown, it includes:

[0036] Step 1, calculating the flow rate data of the mud pump based on the number of pump strokes, includes:

[0037] Q l =A·S·N·Z

[0038] Among them, Q l The values ​​represent the flow rate of the mud pump; A represents the cross-sectional area of ​​the piston; S represents the stroke of the piston within the cylinder; N represents the number of strokes; and Z represents the number of pistons.

[0039] Step 2: Normalize the pressure data and the flow rate data of the mud pump in the monitoring data;

[0040] In this step, each signal is normalized using max-min normalization. Assuming there are n sensors at the hydraulic end of the mud pump, the sample collected by the j-th sensor can be represented as: S' j ={x1,x2,…x m There are a total of m samples. The normalization calculation formula is as follows:

[0041]

[0042] Among them, S′ j For the sample data collected by the j-th sensor, minS' jmaxS' is the minimum value in the original sensor sample set. j S is the maximum value in the original sensor sample set. j For the j-th normalized set of sensor signals, the transformed data will be mapped to the interval [0,1].

[0043] Step 3: Determine the correspondence between flow rate and pressure and the characteristic curve of mud pump operation based on the normalized flow rate data and normalized pressure data.

[0044] The relationship between flow rate and pressure is as follows:

[0045]

[0046] in, A dataset representing flow and pressure. This represents the measured pressure values ​​under different flow conditions. The subscripts 1 to N are the serial numbers under different flow conditions, and t represents the set of measured pressure data.

[0047] The characteristic curve of the mud pump is as follows:

[0048]

[0049] in, A discrete point dataset representing the characteristic curve of a mud pump. The value represents the pressure value fitted to the characteristic curve under different flow conditions. The subscripts 1 to N are the serial numbers under different flow conditions, and r represents the set of pressure data obtained by fitting.

[0050] Step 4: Determine the root mean square error based on the relationship between flow rate and pressure and the characteristic curve of the mud pump operation, including:

[0051] Step 41: Determine the measured pressure value of the mud pump for each flow rate number i in the flow rate and pressure correspondence relationship;

[0052] Step 42: Determine the fitted pressure value corresponding to the characteristic curve under each flow rate number i in the characteristic curve of the mud pump operation;

[0053] Step 43: Calculate the root mean square error of the measured pressure value and the fitted pressure value corresponding to the characteristic curve under the same flow rate using the root mean square formula.

[0054] The root mean square formula is as follows:

[0055]

[0056] In the above formula, RMSE is the root mean square error, N is the total number of samples, and P is the mean square error. i tP represents the measured pressure value of the mud pump under flow rate sequence number i. i r This represents the fitted pressure value corresponding to the characteristic curve under flow rate sequence number i.

[0057] Step 5: Determine the mud pump malfunction based on the root mean square error.

[0058] The calculated RMSE value is compared with a preset threshold. If the RMSE value is greater than the threshold, the mud pump is considered to be faulty, and the steps to analyze and determine the mud pump fault are performed. Otherwise, the mud pump is operating normally.

[0059] In one exemplary embodiment, SFA slow feature analysis is performed on the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals, including:

[0060] Step 1: Determine the transformation function using SFA slow feature analysis;

[0061] Step 2: Use the transformation function to map the vibration data of the mud pump to the feature space, and extract the speed feature signal from the vibration data;

[0062] The fast / slow characteristic signal is:

[0063] S = [S d ,S e ] T

[0064] S d =[S1,S2,…S i ,…,S q ]

[0065] S e =[S q+1 ,S q+2 ,…S i ,…,S m ]

[0066] The above-mentioned fast and slow characteristic signals include m features, S i Let S represent the i-th feature. d S represents the matrix composed of the first q slowest features with the slowest rate of change. e This represents a matrix composed of the remaining mq fast-changing features.

[0067] In this embodiment, SFA analysis is performed on the vibration data of the mud pump in the monitoring data to obtain the fast and slow characteristic signals: (1) For the j-th input sensor signal S j = [x1,x2,…x m ] TThe purpose of SFA is to find a transformation function g(x) = [g1(x), g2(x), ..., g m (x)] T This makes the characteristic F(t) = g(S) j The slowest change.

[0068] (2) The normalized sensor signal S j Perform SVD decomposition on the covariance matrix:

[0069]

[0070] in, Let represent the covariance matrix of the input data, Λ represent the singular value matrix, and U represent the left singular value matrix of Λ.

[0071] Whitening can eliminate cross-correlation between variables. The whitened data can be represented as follows:

[0072] Z = Λ -12 U T S j =QS j

[0073] At this point, SFA is transformed into finding the matrix P = WQ. -1 W = [w1, w2, ..., w m ] T w i Let Q be the weight vector in SFA, defined as: P consists of m orthogonal vectors satisfying S = PZ and P T P = P < ZZ T > t P T =<SS T > t =I, then calculate the difference of the whitened data. Perform a second-order SVD decomposition on the covariance matrix of the difference data, namely:

[0074]

[0075] Therefore, the slow characteristic can be obtained:

[0076] S = PZ = WQ -1 S j

[0077] In this context, the singular values ​​in Ω are arranged in ascending order, representing the rate of change from slow to fast. Smaller singular values ​​correspond to slower features, and vice versa. Based on the rate of change of the features, the acquired features can be divided into two parts:

[0078] S = [Sd ,S e ] T

[0079] S d =[S1,S2,…S i ,…,S q ]

[0080] S e =[S q+1 ,S q+2 ,…S i ,…,S m ]

[0081] The above-mentioned fast and slow characteristic signals include m features, S i Let S represent the i-th feature. d S represents the matrix composed of the first q slowest features with the slowest rate of change. e This represents a matrix composed of the remaining mq fast features with the fastest rate of change. Slow features characterize important operational information of the mud pump when it is not disturbed by external factors, while fast features are considered noise and are therefore omitted.

[0082] In one exemplary embodiment, determining the diagnostic result of the mud pump fault based on the speed characteristic signal includes:

[0083] Step 1: Use XGBoost to perform feature importance analysis on the fast and slow feature signals to obtain a variable importance index change matrix, including:

[0084] Step 11: Generate the corresponding XGBOOST information tree based on the fault sample data, determine the first weight value of the leaf node of the information tree, and obtain the weight vector W at the fault stage. d ;

[0085] In this step, (2) the importance index of variables in the failure phase and the normal phase is calculated using XGBOOST. Taking the failure phase as an example, the failure sample dataset obtained by SFA analysis is X∈R m×n , where m is the number of features and n is the number of sampling points, i.e., the length of the sample.

[0086] The objective function of XGBOOST can be defined as:

[0087]

[0088] Where i is the i-th sample, n is the number of sampling points, k is the k-th tree, l is the loss function, Ω is the penalty term, and f k For tree structure functions, y i It is the measured value of the i-th sample. It is the i-th sample x i Predicted value

[0089] By fitting the residuals of the prediction results from the previous tree to the prediction results from each tree, the resulting model becomes more accurate. The objective function can be transformed into a summation of leaf nodes.

[0090]

[0091] Among them, G j H is the sum of the first-order partial derivatives of the samples contained in leaf node j. j W is the sum of the second-order partial derivatives of the samples contained in leaf node j. j Let λ be the weight of leaf node j, and γ be weighting factors used to control the proportion of the corresponding components. T represents the number of leaf nodes in the tree.

[0092] The splitting of leaf nodes is based on the difference between the objective function value before and after the split. If the gain is positive, it indicates that the split model performs better. The gain calculation formula is as follows:

[0093]

[0094] Where L represents the left subtree and R represents the right subtree. For the information component of the left subtree, For the information component of the right subtree, G represents the amount of information that has not yet been segmented. L G represents the sum of the first-order partial derivatives of the samples in the left tree. R H represents the sum of the first-order partial derivatives of the right-tree samples. L H represents the sum of the second-order partial derivatives of the samples in the left tree. R λ represents the sum of the second-order partial derivatives of the right-tree samples, and γ are weighting factors. The weighting factor λ represents the penalty of L1 regularization, and the weighting factor γ represents the penalty of L2 regularization.

[0095] Taking a fault sample as an example, a corresponding XGBOOST information tree is generated. The weights of the leaf nodes corresponding to each variable can represent the importance of the features. The expression for the weight vector is as follows:

[0096]

[0097] in, This indicates the importance of the m-th feature during the fault phase.

[0098] Step 12: Generate the corresponding XGBOOST information tree based on the normal sample data, determine the second weight value of the leaf nodes of the information tree, and obtain the weight vector W for the normal stage. n ;

[0099] Step 13, based on the weight vector W during the fault stage. d The weight vector W during the normal phase n Obtain the matrix of changes in the importance index of the variables;

[0100] The variable importance index change matrix is ​​as follows:

[0101]

[0102] In the matrix above, D represents the matrix of changes in the importance index of variables. d m This indicates the degree of change of the m-th variable before and after the failure; This indicates the importance of the m-th feature during the fault phase. This indicates the importance of the m-th feature during the normal phase.

[0103] The corresponding weight vector for the normal phase can be obtained:

[0104]

[0105] in, This indicates the importance of the m-th feature during the normal phase.

[0106] By W d With W n A matrix of changes in variable importance indices can be obtained, which measures the degree of change in variables before and after the failure. The calculation results are as follows:

[0107]

[0108] Step 2: Filter the variables according to the variable importance index change matrix to obtain the set of fault variables, including:

[0109] Step 21: Determine whether the element values ​​in the variable importance change matrix are greater than a preset threshold;

[0110] Step 22: If the value of the element is greater than a preset threshold, then the variable corresponding to the element is determined to be a fault variable, and step 23 is executed.

[0111] Step 23: Combine all fault variables to form a fault variable set F = {F1, F2, ..., F...} k}

[0112] Step 3, determine the diagnostic results of the mud pump fault based on the set of fault variables, including:

[0113] Step 31: Extract the time-domain features X∈R for each variable in the fault variable set.k×n ;

[0114] In this step, F1, F2, ..., F k Let F be the set of fault variables. The k-th fault variable can be represented as F. k ={x1,x2,…,x n The dataset contains n samples. Time windows are divided, and temporal features such as mean, standard deviation, root mean square, skewness, and kurtosis are extracted. The calculation formulas for these temporal feature parameters are shown in Table 2 below.

[0115] Table 2

[0116]

[0117] Step 32: Perform SVM diagnosis based on the time-domain feature signal to determine the diagnosis result of the single variable; the time-domain feature obtained from step 31 can be described as X∈R k×n , where n is the number of samples and k is the dimension of each sample. The hyperplane defined by SVM can be represented as:

[0118] f(x) = w·x + b

[0119] Where w is the optimal classification surface weight coefficient vector, x is the training sample, and b is the bias.

[0120] The optimal hyperplane can be transformed into:

[0121]

[0122]

[0123] Where, ξ i This represents a non-negative relaxation variable introduced under constraints, and C is a trade-off between training error and confidence range, i.e., a penalty factor.

[0124] The optimal classification function can be defined as:

[0125] f(x) = sgn[(w * ·x)+b * ]=sgn[α * y i (x i ·x)+b * ]

[0126] Among them, w * This represents the optimal solution of the final calculated optimal classification surface weight coefficient vector, α. * For Lagrange multipliers, y i (x i ·x) is the kernel function, b * This is the optimal solution for the bias.

[0127] Step 33: Use DS evidence to fuse the diagnostic results of each variable to obtain the final fault diagnosis result.

[0128] In this step, the training results of a single dataset are used as evidence, and fusion is performed using DS (Discretionary Data Structure) evidence argumentation. The DS synthesis rules are as follows:

[0129]

[0130] Where k describes the degree of conflict of the evidence, m1 and m2 are the corresponding basic probability assignment functions, E and F are the constituent elements of the corresponding events, and m is the evidence body fused using Dempster's rule.

[0131] To utilize conflicting information, the probability of evidence conflict is weighted according to credibility and allocated to each proposition. Therefore, the fault diagnosis model can be written as:

[0132] m(φ)=0

[0133]

[0134] Among them, A i For the i-th focal element, m i Let be the basic probability assignment function for the i-th focal element, and n be the total number of propositions.

[0135] In this embodiment, sensors deployed on the equipment collect data from various measuring points. Data is transmitted from the field equipment to the server via data acquisition components, data reading components, and remote data transmission, ultimately integrating sensor parameters into the drilling equipment fault diagnosis system. Characteristic curve data is introduced and compared with actual flow-pressure data and flow-pressure characteristic curve data to detect whether the mud pump has malfunctioned. If a malfunction occurs, slow feature analysis is used to extract feature information from multi-sensor signals. XGBoost is used to compare the importance of features before and after the malfunction, filtering out feature variables with significant changes and summarizing them into a fault variable set. The final filtered fault variables are then used for individual information source diagnosis via SVM, and decision fusion is performed using DS evidence argumentation to arrive at a fault diagnosis conclusion. This method has excellent feature extraction capabilities under complex operating conditions, possesses a more reliable parameter fusion method, and exhibits superior accuracy and robustness.

[0136] This invention provides a drilling equipment fault diagnosis device, such as... Figure 4 As shown, the device includes a memory 400 and a processor 410; the memory is used to store a program for diagnosing drilling equipment faults, and the processor is used to read and execute the program for diagnosing drilling equipment faults, and to execute the method described in any of the above embodiments.

[0137] This invention also provides a computer-readable storage medium storing a data processing program, which is executed by a processor using the drilling equipment fault diagnosis method described in any of the above embodiments.

[0138] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A method for diagnosing faults in drilling equipment, characterized in that, The method includes: The equipment collects relevant monitoring data in real time through sensors; the monitoring data includes: mud pump vibration data and mud pump pressure data; The mud pump malfunction was determined based on the mud pump pressure data in the monitoring data. Slow feature analysis was performed on the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals; The diagnosis result of the mud pump fault is determined based on the speed characteristic signal; The step of determining a mud pump malfunction based on the mud pump pressure data in the monitoring data includes: The flow rate data of the mud pump is calculated based on the number of pump strokes; the pressure data and flow rate data of the mud pump in the monitoring data are normalized; the correspondence between flow rate and pressure and the characteristic curve of mud pump operation are determined based on the normalized flow rate data and normalized pressure data; the root mean square error is determined based on the measured pressure data of the flow rate and pressure correspondence and the fitted pressure data of the characteristic curve; and the mud pump failure is determined based on the root mean square error. The step of determining the diagnostic result of the mud pump fault based on the fast and slow characteristic signals includes: using XGBOOST to perform feature importance analysis on the fast and slow characteristic signals to obtain a variable importance index change matrix; filtering variables based on the variable importance index change matrix to obtain a fault variable set; and determining the diagnostic result of the mud pump fault based on the fault variable set. The step of using XGBOOST to perform feature importance analysis on the fast and slow feature signals to obtain a variable importance index change matrix includes: generating a corresponding XGBOOST information tree based on fault sample data, determining the first weight value of the leaf nodes of the information tree, and obtaining the weight vector during the fault stage; generating a corresponding XGBOOST information tree based on normal sample data, determining the second weight value of the leaf nodes of the information tree, and obtaining the weight vector during the normal stage; and obtaining the variable importance index change matrix based on the weight vector during the fault stage and the weight vector during the normal stage.

2. The drilling equipment fault diagnosis method according to claim 1, characterized in that, The relationship between flow rate and pressure is as follows: ； in, A dataset representing flow and pressure. This represents the measured pressure values ​​under different flow conditions. The subscripts 1 to N are the serial numbers under different flow conditions, and t represents the set of measured pressure data.

3. The drilling equipment fault diagnosis method according to claim 1, characterized in that, The characteristic curve of the mud pump is as follows: ； in, A discrete point dataset representing the characteristic curve of a mud pump. This represents the pressure value fitted to the characteristic curve under different flow conditions. The subscripts 1 to N are the serial numbers under different flow conditions, and r represents the set of fitted pressure data.

4. The drilling equipment fault diagnosis method according to claim 1, characterized in that, The determination of the root mean square error based on the measured pressure data of the correspondence between flow rate and pressure and the fitted pressure data of the characteristic curve includes: Determine the measured pressure value of the mud pump for each flow rate number i in the flow rate-pressure correspondence; Determine the fitted pressure value corresponding to the characteristic curve under each flow rate number i in the characteristic curve of the mud pump; The root mean square error between the measured pressure value and the fitted pressure value corresponding to the characteristic curve under the same flow rate sequence number is calculated using the root mean square formula. The root mean square formula is as follows: ； In the above formula, RMSE is the root mean square error, and N is the total number of samples. The measured pressure value of the mud pump under flow rate sequence number i. This represents the fitted pressure value corresponding to the characteristic curve under flow rate sequence number i.

5. The drilling equipment fault diagnosis method according to claim 1, characterized in that, The slow feature analysis of the vibration data of the mud pump in the monitoring data to obtain fast and slow feature signals includes: Determine the transition function using slow feature analysis; The vibration data of the mud pump is mapped to a feature space using the transformation function, and the speed feature signals in the vibration data are extracted. The fast / slow characteristic signal is: ； ； ； The above-mentioned fast and slow characteristic signals include m features. Represents the i-th feature. This represents a matrix composed of the first q slowest features with the slowest rate of change. This represents a matrix composed of the remaining mq fast-changing features.

6. The drilling equipment fault diagnosis method according to claim 1, characterized in that, The change matrix of the variable importance index is as follows: ； In the matrix above, D represents the matrix of changes in the importance index of variables. d m This indicates the degree of change of the m-th variable before and after the failure; This is the weight vector during the fault phase. , This indicates the importance of the m-th feature during the fault phase. This is the weight vector during the normal phase. , This indicates the importance of the m-th feature during the normal phase.

7. The drilling equipment fault diagnosis method according to claim 6, characterized in that, The set of fault variables obtained by filtering variables based on the variable importance index change matrix includes: Determine whether the element values ​​in the variable importance index change matrix are greater than a preset threshold; If the value of the element is greater than a preset threshold, then the variable corresponding to the element is determined to be a fault variable; Combine all fault variables to form a fault variable set. .

8. The drilling equipment fault diagnosis method according to claim 7, characterized in that, The step of determining the diagnostic result of the mud pump fault based on the set of fault variables includes: Extract the time-domain features of each variable in the set of fault variables. Where n is the number of samples, and k is the dimension of each sample; SVM diagnosis is performed based on the time-domain features to determine the diagnostic result for each variable; The diagnostic results of each variable are integrated using DS evidence to obtain the final fault diagnosis result.

9. A fault diagnosis device for drilling equipment, characterized in that, The device includes a memory and a processor; the memory is used to store a program for diagnosing drilling equipment faults, and the processor is used to read and execute the program for diagnosing drilling equipment faults, and to execute the method according to any one of claims 1-8.

10. A computer-readable storage medium storing a data processing program, the data processing program being executed by a processor according to any one of claims 1-8, the drilling equipment fault diagnosis method.

Images