Rotary drilling machine fault early warning method under multi-source data analysis

Abstract

The invention discloses a rotary drilling machine fault early warning method under multi-source data analysis, and relates to the technical field of fault early warning management.The method comprises the steps that multi-source sensing data of a target rotary drilling machine are collected in real time based on a multi-source sensing module, and after the data are preprocessed, terrain influence features, hydraulic system features and ground environment features are extracted; and performing dynamic modeling and health assessment based on the multivariate feature information to obtain a health state assessment result of the target rotary drilling machine, and performing equipment fault prediction through a multivariate fault prediction mechanism to generate a fault early warning signal. The technical problems that an existing rotary drilling machine fault early warning method depends on single data and a static model, the dynamic change of equipment cannot be reflected in time, and the accuracy of fault prediction is low are solved, and the purposes of performing rotary drilling machine fault prediction and health management by fusing multi-source sensing data and a self-adaptive updating mechanism and improving the safety of the rotary drilling machine are achieved. And the real-time performance and the accuracy of the fault early warning of the rotary drilling machine are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault early warning management, and specifically relates to a fault early warning method for a rotary drilling rig under multi-source data analysis. Background Art

[0002] As an efficient engineering construction machinery, a rotary drilling rig is widely used in infrastructure construction, mine exploitation, underground engineering and other fields. However, during the operation process of the rotary drilling rig, it often faces challenges of harsh environmental conditions, such as irregular terrain, extreme climate conditions or overloading of the equipment. These factors may lead to excessive wear, fatigue damage of equipment components and abnormalities in the hydraulic system, affecting the performance and safety of the equipment. Therefore, ensuring the reliable operation of the rotary drilling rig and predicting equipment faults in advance have become the key to improving equipment efficiency, reducing maintenance costs and ensuring construction safety.

[0003] Traditional fault monitoring and early warning methods for rotary drilling rigs mainly rely on the monitoring of single-device sensing data, such as temperature, pressure, vibration, etc. However, in a complex environment, especially when operating on irregular terrain (such as slopes, soft soil, water accumulation, etc.), these methods cannot comprehensively and accurately capture the true working state and potential faults of the rotary drilling rig to effectively predict the faults of the rotary drilling rig and manage its health. Summary of the Invention

[0004] The present application provides a fault early warning method for a rotary drilling rig under multi-source data analysis, which is used to solve the technical problems that the existing fault early warning methods for rotary drilling rigs rely on single data and static models, cannot reflect the dynamic changes of the equipment in time, and have low accuracy in fault prediction.

[0005] The present application provides a fault early warning method for a rotary drilling rig under multi-source data analysis. The method includes: based on a multi-source sensing module, real-time collecting multi-source sensing data of a target rotary drilling rig, where the multi-source sensing data includes hydraulic system data, equipment sensing data, terrain sensing data and historical operation and maintenance data; after preprocessing the multi-source sensing data, performing multi-source feature information extraction, where the multi-source feature information includes terrain influence features, hydraulic system features and ground environment features; based on the multi-source feature information, performing dynamic modeling and health assessment to obtain a health status assessment result of the target rotary drilling rig; according to the health status assessment result, through a multi-source fault prediction mechanism, performing equipment fault prediction to generate a fault early warning signal.

[0006] One or more technical solutions provided in the present application have at least the following technical effects or advantages: The rotary drilling rig fault warning method under multi-source data analysis provided by this application relates to the technical field of fault warning management. It collects multi-source sensing data such as the hydraulic pressure, sensors, and terrain of the rotary drilling rig in real time through a multi-source sensing module, conducts preprocessing and feature extraction, and performs dynamic modeling and health assessment based on the extracted features. According to the health status assessment results, a multi-source fault prediction mechanism is used to predict equipment faults and generate fault warning signals, solving the technical problems that the existing rotary drilling rig fault warning methods rely on single data and static models, cannot reflect the dynamic changes of the equipment in a timely manner, and have low accuracy in fault prediction. It realizes the technical effect of performing rotary drilling rig fault prediction and health management by integrating multi-source sensing data and an adaptive update mechanism, improving the timeliness and accuracy of rotary drilling rig fault warning. Description of the Drawings

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0008] Figure 1 It is a schematic flowchart of the rotary drilling rig fault warning method under multi-source data analysis provided by the embodiments of this application; Figure 2 It is a schematic flowchart of dynamic modeling and health assessment in the rotary drilling rig fault warning method under multi-source data analysis provided by the embodiments of this application. Detailed Embodiments

[0009] This application provides a rotary drilling rig fault warning method under multi-source data analysis to solve the technical problems that the existing rotary drilling rig fault warning methods rely on single data and static models, cannot reflect the dynamic changes of the equipment in a timely manner, and have low accuracy in fault prediction.

[0010] The following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the drawings in the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, not all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of this application without creative efforts belong to the scope of protection of this application.

[0011] It should be noted that the terms "first", "second", etc. in the description of this application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0012] Embodiment 1, as Figure 1 shown, the present application provides a method for predicting the faults of a rotary drilling rig under multi-source data analysis. The method includes: P10: Based on a multi-source sensing module, collect multi-source sensing data of the target rotary drilling rig in real time. The multi-source sensing data includes hydraulic system data, equipment sensing data, terrain sensing data, and historical operation and maintenance data.

[0013] Specifically, deploy a multi-source sensing module to collect multi-source sensing data of the target rotary drilling rig in real time. These data provide basic information for the status monitoring and fault warning of the rotary drilling rig, ensuring that the system can comprehensively and real-time track various operating indicators of the equipment.

[0014] In this step, the rotary drilling rig is equipped with a multi-source sensing module. The multi-source sensing module includes multiple sensor modules for different data sources, which are used to collect various types of data in real time, and collect various working status data of the rotary drilling rig through the multi-source sensing module in real time. Specifically, it includes: By installing hydraulic sensors (such as pressure sensors and flow sensors), monitor key parameters such as the pressure and flow of hydraulic oil in real time, and obtain hydraulic system data. These data can help judge the load and working efficiency of the hydraulic system, and timely detect possible leakage or wear problems in the system.

[0015] In addition to the hydraulic system data, the multi-source sensing data also includes equipment sensing data, which involves the real-time operating conditions of other core components of the rotary drilling rig. For example, sensors can monitor engine temperature, vibration, oil temperature, rotation speed, etc. Through the data of these sensors, the operating status of the equipment can be comprehensively understood, ensuring that the mechanical components do not operate overloaded, and timely identifying potential fault hazards.

[0016] In addition, terrain sensing data is particularly crucial, especially when operating on irregular terrains. This data can include information such as slope, ground hardness, soil moisture, rock distribution, etc., helping the system comprehensively understand the changes in the operating environment. Terrain sensors can obtain accurate terrain information through LiDAR, ground pressure sensors, or ultrasonic sensors. These data can help the system judge the impact of ground conditions on the rotary drilling rig, such as whether the ground is soft, whether there is water accumulation, or whether there are obstacles such as large rocks, which may affect the operating stability of the rotary drilling rig.

[0017] Meanwhile, historical operation and maintenance data provides a background and reference for existing operations. By obtaining the historical operation and maintenance data of the target rotary drilling rig, including past fault records, maintenance history, component replacement situations, etc., it can help the model identify potential problem trends of the equipment. Historical data can also be provided to machine learning models for training, thereby improving the accuracy and reliability of fault prediction.

[0018] To achieve efficient data collection and fusion, the multi-source sensing module needs to have the capabilities of high-frequency sampling and low-latency transmission, for example, by means of an embedded computing unit and wireless transmission technologies (such as 5G or Wi-Fi). After data transmission, it is aggregated to a central processing unit or an edge computing node for further processing and analysis. Through such real-time collection and transmission of multi-source data, the operating state of the rotary drilling rig and external environmental factors can be monitored in real time, providing accurate data to support decision-making.

[0019] P20: After preprocessing the multi-source sensing data, perform multi-feature information extraction. The multi-feature information includes terrain impact features, hydraulic system features, and ground environment features.

[0020] Furthermore, step P20 of the embodiment of the present application further includes: P21: Deploy multiple edge computing nodes in the control system of the target rotary drilling rig. The edge computing nodes are communicatively connected to the multi-source sensing module and can directly obtain data from the sensors; P22: The edge computing nodes receive the multi-source sensing data and perform data preprocessing at each node to generate standardized multi-source data; P23: On the edge computing nodes, perform multi-feature information extraction on the standardized multi-source data to obtain terrain impact features, hydraulic system features, and ground environment features, and convert the terrain impact features, hydraulic system features, and ground environment features into a multi-source feature matrix.

[0021] It should be understood that the raw data obtained from the multi-source sensing module is preprocessed, and then by extracting multi-source feature information, it provides key information for subsequent dynamic modeling and health assessment. This step is an important link in multi-source data analysis, covering operations such as data cleaning, normalization processing, and feature extraction to ensure data quality and analysis accuracy.

[0022] First, in the control system of the target rotary drilling rig, multiple edge computing nodes are deployed, and these nodes are communicatively connected to the multi-source sensing module. The role of the edge computing nodes is to transfer data processing from the central server to the device site, which can reduce data transmission latency and improve processing efficiency. The edge computing nodes can directly obtain data from the sensors and perform data preprocessing locally. Different from traditional cloud computing, edge computing can achieve faster data response, especially in complex operating environments, and can process and respond to fault signals in real time.

[0023] After receiving the multi-source sensing data on the edge computing nodes, data preprocessing is first carried out. The purpose of preprocessing is to clean and transform the raw data to make it meet the requirements of subsequent analysis. Common preprocessing techniques include denoising, filling missing values, outlier detection, and data normalization. Through normalization processing, data from different sources are transformed into the same scale range, thereby eliminating the dimensional differences between sensors and ensuring the comparability of each data type during analysis. Data preprocessing can also include time synchronization to ensure that data from different sensors is consistent in the time dimension.

[0024] After preprocessing, next, multi-source feature information is extracted from the normalized multi-source data on the edge computing nodes, and multi-source features related to the operation safety and stability of the rotary drilling rig are extracted. Specifically, the extracted multi-source feature information includes the following categories: Terrain influence features, including terrain slope, ground hardness, wetness degree, etc. Terrain features are crucial for aspects such as the stability of the equipment and the load uniformity. Especially in irregular terrains (such as large slopes, soft soil, water accumulation, etc.), it may cause the equipment to have unstable center of gravity, reduced operation efficiency, or uneven load on the hydraulic system.

[0025] Hydraulic system features, including hydraulic pressure, flow rate, oil temperature, etc., which can reflect the load situation and health status of the hydraulic system. If abnormalities occur in the hydraulic system (such as too high pressure, overheated oil temperature, etc.), it may lead to a decline in equipment performance and even failures.

[0026] Ground environment features, including environmental factors such as ground temperature, humidity, wind speed, etc., which describe the working environment where the rotary drilling rig is located. These environmental conditions may affect the operation stability of the equipment. For example, a slippery ground may cause tire slippage or increase the load on the hydraulic system, while too high a temperature may cause the hydraulic oil temperature to rise, affecting the system efficiency.

[0027] The extracted feature data will be converted into a multi-source feature matrix, which pools information from different data sources and feature dimensions, providing comprehensive inputs for subsequent dynamic modeling, health assessment, and fault prediction. Each dimension in the feature matrix represents an important device state or environmental feature, which will help identify potential fault patterns and device health risks in subsequent analysis.

[0028] By performing data preprocessing and feature extraction on edge computing nodes, it is possible to ensure that data is processed closest to the device, reducing latency and improving the response speed and accuracy of the entire early warning system. This process not only improves the real-time performance of the system but also ensures the effective fusion of multi-source data, ultimately providing high-quality input data for the health state assessment and fault prediction of the rotary drilling rig.

[0029] P30: Based on the multi-source feature information, perform dynamic modeling and health assessment to obtain the health state assessment result of the target rotary drilling rig.

[0030] Furthermore, as Figure 2 shown, step P30 of the embodiment of the present application further includes: P31: Perform multi-physics field modeling based on the key components of the target rotary drilling rig to establish a preliminary dynamics model; P32: Extract multi-source feature information through the multi-source feature matrix, dynamically empower the preliminary dynamics model to generate a dynamic working model; P33: According to the multi-source sensing data, use the dynamic working model to perform operation path simulation, and collect simulation data for multi-level real-time health assessment to generate the health state assessment result of the target rotary drilling rig.

[0031] Optionally, based on the multi-source feature information extracted from the multi-source sensing data, comprehensively evaluate the health state of the target rotary drilling rig through means such as multi-physics field modeling technology, dynamic empowerment, and real-time simulation to obtain the health state assessment result of the target rotary drilling rig.

[0032] First, by analyzing the key components of the rotary drilling rig (such as the hydraulic system, power system, transmission system, etc.), a preliminary dynamics model is established. This model mainly involves the kinematics and dynamics behavior of the equipment and the interaction between various components. Multi-physics field modeling refers to a model that comprehensively considers multiple physical fields (such as mechanics, electricity, hydraulics, thermotics, etc.), which can reflect the real behavior of the equipment under complex working conditions. For example, when the rotary drilling rig is operating, the interaction between the hydraulic system and the power system will affect the overall operation efficiency, and the terrain factor will also affect the movement trajectory of the machine. Therefore, by constructing a preliminary multi-physics field model, the dynamic changes of the equipment under different working conditions can be accurately captured.

[0033] The multi-physics field modeling can use finite element analysis (FEA) or other numerical simulation methods to describe and predict the responses of key components of the rotary drilling rig. For example, when modeling the hydraulic system, factors such as pressure, flow rate, and temperature that affect hydraulic components can be considered, while when modeling the mechanical structure, its deformation, stress, and fatigue behavior under different working conditions need to be considered. Through these physical field models, the dynamic characteristics and working states of the rotary drilling rig can be initially understood.

[0034] Next, the preliminary dynamic model is dynamically empowered through the multi-source feature information extracted from the multi-source feature matrix. Dynamic empowerment refers to adjusting and optimizing the dynamic model according to real-time sensing data and changes in working conditions, enabling it to more accurately reflect the behavior of the rotary drilling rig during actual operation. Through the empowerment process, environmental factors such as hydraulic load, terrain influence, and equipment tilt can be integrated into the model, thus generating a dynamic working model.

[0035] The dynamic working model can not only take into account the basic dynamic characteristics of the rotary drilling rig but also adapt to changes in the operating environment. For example, in irregular terrains (such as areas with large slopes or soft soil), the working state of the rotary drilling rig will be different, so the dynamic model needs to be adjusted according to the real-time collected terrain features and environmental data. Through dynamic empowerment, the working model can be adaptively adjusted according to real-time data changes, ensuring that the model always remains consistent with the actual working state of the equipment.

[0036] After the dynamic working model is established, it is used for operation path simulation. The purpose of operation path simulation is to predict the performance and health status of the equipment in different scenarios by simulating the movement trajectories of the rotary drilling rig in different terrains and working environments. For example, when the rotary drilling rig is working in an area with a large slope, the simulation can help us determine whether the equipment may tilt or the hydraulic system may be overloaded due to unstable center of gravity.

[0037] During the simulation process, various data obtained from sensors and the model are continuously collected, and through these data, a multi-level real-time health assessment of multiple systems of the rotary drilling rig is carried out. This includes individual health assessments of the power system, hydraulic system, control system, and mechanical components, and a comprehensive health assessment result is generated based on the operating conditions of each part. For example, the hydraulic system may experience excessive wear due to uneven load or too high oil temperature, while the power system may be fatigued and damaged due to long-term high-load operation. Through multi-level assessment, a detailed health status report can be generated for each system, and finally, a health status assessment result of the entire rotary drilling rig is synthesized.

[0038] The health assessment results can not only reflect the current operating status of the equipment, but also predict the possible future failure risks, providing a decision-making basis for maintenance personnel. Through this health assessment method based on multi-physical field modeling and dynamic empowerment, the maintenance management of the rotary drilling rig will become more accurate, reducing the problem of failure warning delay caused by the lack of real-time feedback in traditional methods.

[0039] Furthermore, step P32 of the embodiment of the present application further includes: P32-1: Extract terrain influence features, hydraulic system features, and ground environment features based on the multi-source feature matrix; P32-2: Quantify the variable factors according to the terrain influence features, hydraulic system features, and ground environment features to generate multiple model parameter adjustment coefficients, including terrain adjustment coefficient, soil adjustment coefficient, load adjustment coefficient, and hydraulic adjustment coefficient; P32-3: Use the multiple model parameter adjustment coefficients to dynamically empower the preliminary dynamic model to generate the dynamic working model.

[0040] In a possible embodiment of the present application, the dynamic empowerment process is further refined, and the feature information extracted from multi-source data provides more adjustment bases for the dynamic model, so that the preliminary dynamic model can more accurately reflect the working state of the rotary drilling rig under different environmental conditions.

[0041] First, based on the multi-source feature matrix, select the feature information closely related to the operation of the rotary drilling rig, including terrain influence features, hydraulic system features, and ground environment features. The terrain influence features include information such as terrain slope, soil type, and ground hardness, which will directly affect the stability and working efficiency of the rotary drilling rig. For example, when operating on a steep slope or soft soil, the force condition of the rotary drilling rig will be different from that on flat terrain; the hydraulic system features mainly include data such as the pressure, flow rate, and temperature of the hydraulic system, which can reflect the working state of the hydraulic system. Especially under heavy load and unstable terrain, the pressure of the hydraulic system may change, leading to system overload or damage; the ground environment features include soil humidity, ground hardness, etc., which have a significant impact on the crawler walking performance and operation efficiency of the rotary drilling rig. By extracting this feature information, it is ensured that the working environment of the rotary drilling rig can be fully reflected in the dynamic model, making the subsequent health assessment more accurate.

[0042] After the feature data is extracted, the next step is to quantify these features as variable factors. That is, each feature is mathematically modeled with the relevant parameters in the operation model of the rotary drilling rig and converted into quantifiable values. These quantified features will become the model parameter adjustment coefficients for dynamically adjusting the parameters in the dynamic model. Specifically, it includes: The terrain adjustment coefficient can quantify the impacts of factors such as the slope, hardness, and flatness of the terrain on the stability and load of the rotary drilling rig based on terrain influence characteristics. For example, when operating on a steep slope, the terrain adjustment coefficient will increase accordingly to reflect the potential risk of unstable center of gravity; the soil adjustment coefficient can quantify the impact of the softness and hardness of the soil on the crawler walking performance of the rotary drilling rig according to the soil type and humidity in the ground environment characteristics. When the soil is relatively soft, the soil adjustment coefficient may be relatively high, indicating that the traction force of the rotary drilling rig will be affected; the load adjustment coefficient refers to the impact of changes in the load of the hydraulic system on the working state. For example, when the hydraulic system bears a large load, the load adjustment coefficient will increase, reflecting the increase in the pressure demand of the hydraulic system; the hydraulic adjustment coefficient can quantify the impact of changes in the working pressure, flow rate, etc. of the hydraulic system on the performance of the rotary drilling rig based on the characteristics of the hydraulic system. When the load is heavy or the environment is unstable, the hydraulic adjustment coefficient will increase, indicating that the hydraulic system may need to bear greater pressure.

[0043] These adjustment coefficients can be derived by analyzing historical data, actual operation conditions, and simulation models, and endow the subsequent dynamic model with more realistic and accurate adjustment factors.

[0044] Next, use the multiple model parameter adjustment coefficients generated above to dynamically empower the preliminary dynamic model. The process of dynamic empowerment is to apply these adjustment coefficients to the preliminary dynamic model to adjust the various parameters of the model to make it more conform to the actual working state. For example: for the terrain adjustment coefficient, the stability and load distribution of the rotary drilling rig can be adjusted in the model to reflect the impact of the terrain on the operation. For the hydraulic adjustment coefficient, the working pressure and flow rate of the hydraulic system can be adjusted to ensure the stable operation of the hydraulic system under different environments.

[0045] The core objective of dynamic empowerment is to enable the preliminary dynamic model to reflect the dynamic behavior of the rotary drilling rig in various working environments in real time through fine-tuning of the model parameters, and then generate a dynamic working model that can adapt to different working conditions. This model can accurately predict the various performances of the rotary drilling rig, such as the load changes of the hydraulic system, the stress distribution of the equipment structure, etc., so as to provide reliable data support for subsequent health assessment, fault prediction, and maintenance decision-making.

[0046] Furthermore, step P33 of the embodiment of the present application further includes: P33-1: Based on the multi-source sensing data, perform operation path planning for the target rotary drilling rig to generate the optimal planned path; P33-2: According to the optimal planned path, use the dynamic working model to simulate the operation path, and during the simulation process, collect multi-component state data sets according to the equipment component levels respectively; P33-3: Based on the multi-component state data sets, conduct multi-level real-time health assessment on the target rotary drilling rig, including power system health assessment, hydraulic system health assessment, control system health assessment, and mechanical component health assessment, and fuse the multi-level assessment results to generate the health status assessment result of the target rotary drilling rig.

[0047] Specifically, based on the collected multi-source sensing data, simulate the operation path through the dynamic working model, and conduct multi-level real-time health assessment on the target rotary drilling rig. This process aims to accurately assess the health status of the rotary drilling rig from multiple dimensions (including power system, hydraulic system, control system, and mechanical components), and generate the final health status assessment result, providing a scientific basis for fault warning and maintenance decision-making.

[0048] First, utilize the real-time terrain, environment, and equipment state data collected from the multi-source sensing module, and through the intelligent operation path planning algorithm, formulate the optimal planned path for the rotary drilling rig. This path planning not only considers factors such as terrain difficulty, slope, soil conditions, etc., but also combines internal parameters such as the load capacity of the rotary drilling rig and the performance of the hydraulic system to ensure that the rotary drilling rig can perform at its best in the most suitable route and operation environment. The goal of the planned path is to maximize work efficiency while avoiding overloading and unstable operation of the equipment, reducing the wear and failure risks of the equipment.

[0049] After determining the optimal planned path, the next step is to use the dynamic working model to simulate the operation path. By simulating the working states of each component of the rotary drilling rig under different paths, collect the multi-component state data sets at each stage. These data include but are not limited to: the working pressure of the hydraulic system, the power demand of the power system, the traction force change of the crawler, etc. During the simulation process, the dynamic working model will adjust the states of each component according to the real-time collected sensing data to reflect the working performance under different operation conditions. Through this process, the performance and load conditions of each component during the path simulation can be grasped in real time, providing accurate data for the subsequent health assessment.

[0050] Next, according to the collected multi-component state data sets, we conduct multi-level real-time health assessment on each system and component of the rotary drilling rig. This health assessment includes the following aspects: Power system health assessment: Based on parameters such as the power output, engine speed, and load ratio of the power system, it assesses whether the engine and transmission system are within the normal operating range and identifies possible mechanical failures or performance degradation. Hydraulic system health assessment: Based on the pressure, flow rate, and temperature data of the hydraulic system, it evaluates the operating status of the hydraulic pump, hydraulic oil pipelines, and valves to determine whether there are problems such as leaks, blockages, or overloads. Control system health assessment: Based on the feedback data of the control system (such as sensor signals, actuator responses, etc.), it assesses whether the control unit responds accurately and effectively to instructions and whether there are faults or abnormalities. Mechanical component health assessment: By monitoring the working status of the mechanical components of the rotary drilling rig, such as the crawler, slewing platform, drill pipe, etc., it determines whether there is excessive wear or structural damage.

[0051] Furthermore, through the fusion of multi-level assessment results, the assessment results of each subsystem (such as the power system, hydraulic system, control system, and mechanical components) are integrated to generate the final health status assessment result of the target rotary drilling rig. This fusion process can be achieved through algorithms such as weighted average and Bayesian fusion to ensure that the importance of each subsystem is reasonably considered. The final health assessment result can reflect the overall health level of the rotary drilling rig in the current working environment and state. This process not only improves the accuracy of fault warning but also provides strong data support for the maintenance and optimization of the rotary drilling rig.

[0052] Furthermore, step P30 of the embodiment of the present application further includes: The dynamic working model is embedded with an adaptive update mechanism. Based on this adaptive update mechanism, it receives continuously updated multi-source feature information in real time and performs incremental learning and dynamic update on the dynamic working model.

[0053] It should be understood that the dynamic working model is embedded with an adaptive update mechanism, which allows the dynamic working model to perform incremental learning and dynamic update in real time according to the changing operating environment and equipment status information, thereby improving the prediction accuracy and response ability of the model.

[0054] Since rotary drilling rigs often work on irregular terrains (such as steep slopes, soft soil, waterlogged areas, or rocky areas), the environmental conditions and operation paths may change drastically. The adaptive update mechanism can adjust the model according to real-time multi-source data, enabling the model to always maintain accuracy and robustness in a dynamic environment, thereby enhancing the response ability of the rotary drilling rig in complex environments.

[0055] First, the adaptive update mechanism of the dynamic working model continuously receives new data and multi-source feature information from multiple sensors. These feature information not only includes conventional data such as the working state of the hydraulic system and the health status of mechanical components, but also includes terrain influence features (such as slope changes, soil softness) and ground environment features (such as external factors like climate, humidity). As the operating environment and tasks of the rotary drilling rig change, the sensor data will also change continuously, resulting in dynamic changes in the operating load and working conditions faced by the equipment. Therefore, by continuously receiving this updated information, the dynamic working model can be adjusted according to the changes in actual operating conditions to maintain an accurate prediction of the equipment status.

[0056] Moreover, the adaptive update mechanism of the dynamic working model continuously optimizes the model parameters through incremental learning technology. Incremental learning can ensure that the model gradually learns when receiving new multi-source feature data without having to retrain the entire model from scratch. This method greatly improves the computational efficiency, especially when the equipment is in a large-scale operation state, it can avoid frequent training of all data and reduce the consumption of computing resources.

[0057] Exemplarily, after receiving a new data set each time, the incremental learning mechanism fine-tunes the prediction parameters of the model. For example, if the hydraulic system shows different pressure and flow characteristics in a new operating environment, the incremental learning mechanism will automatically adjust the hydraulic adjustment coefficient so that the model can reflect this change in real time. In addition, incremental learning can also correct inaccurate predictions in historical data through a certain feedback mechanism, thereby continuously improving the accuracy and reliability of predictions. This process not only improves the accuracy and flexibility of the model, but also ensures that the rotary drilling rig can always adapt to various changes in complex operating environments, improving the safety and efficiency of operations.

[0058] P40: According to the health status assessment result, through a multi-source fault prediction mechanism, conduct equipment fault prediction and generate a fault warning signal.

[0059] Furthermore, step P40 of the embodiment of the present application further includes: P41: Generate multiple related prediction tasks according to the equipment component level and establish a multi-source fault prediction mechanism based on this; P42: Through the multi-source fault prediction mechanism, simultaneously conduct fault prediction for the multiple related prediction tasks to generate a multi-source fault prediction result; P43: Weight and fuse the multi-source fault prediction results to generate the fault warning signal.

[0060] Optionally, based on the health status assessment result of the rotary drilling rig, conduct equipment fault prediction through a multi-source fault prediction mechanism and finally generate a fault warning signal.

[0061] First, based on the health assessment results of the main components of the rotary drilling rig, such as the hydraulic system, power system, mechanical components, control system, etc., the equipment is decomposed into multiple component levels. Each component level represents an independent prediction task. For example, the hydraulic system may require an independent fault prediction task, while the mechanical components and the power system may be another task. These prediction tasks will respectively evaluate the health status of the corresponding components and identify potential faults. Through this hierarchical modeling, the independent fault modes and characteristics of each component can be captured with finer granularity.

[0062] On this basis, a multi-fault prediction mechanism is established. This mechanism can support multiple tasks to be carried out simultaneously, and each task adopts different prediction models or algorithms according to the characteristics of specific components, such as support vector machines, decision trees, neural networks, etc. At this time, the prediction mechanism will process the data streams from different components and generate independent prediction results for each task.

[0063] With the support of the multi-fault prediction mechanism, the health assessment data of all components will be input into multiple relevant prediction tasks simultaneously. These tasks include: hydraulic system fault prediction, using hydraulic sensor data (such as pressure, flow rate, temperature, etc.) to predict the health of the hydraulic system and identify whether there are potential problems such as leaks, wear, or abnormal pressure. Mechanical component fault prediction, based on the vibration sensor data of mechanical components (such as accelerometer data), to predict whether there are faults such as wear, looseness, fatigue, etc. Power system fault prediction, using engine data (such as load, speed, temperature, etc.) to conduct fault prediction and determine whether the engine is in an abnormal working state. Control system fault prediction, according to the working state of the control system, to monitor whether there are software faults, sensor failures, or communication errors, etc.

[0064] After the distribution of the prediction tasks is completed, the next step is to synchronously process multiple prediction tasks through the multi-fault prediction mechanism. Each prediction task will process based on the multi-source feature data of the relevant components, and use technologies such as machine learning models or time series analysis (such as long short-term memory network LSTM) to conduct fault prediction on different components of the equipment. The core of the multi-fault prediction mechanism lies in its ability to process multiple tasks simultaneously, and each task outputs the fault probability or risk assessment value for a specific component.

[0065] For example, the fault prediction of the hydraulic system may rely on the characteristics of the hydraulic system (such as sensing data of pressure, flow rate, temperature, etc.), while the fault prediction of the control system may need to combine the characteristics of the control system (such as current, voltage, equipment sensing data, etc.). These different data sources will be input into their respective prediction models simultaneously, and the fault risk of each component will be calculated through the models.

[0066] After obtaining the fault prediction results at multiple component levels, in order to obtain a more accurate overall fault prediction, the respective prediction results are weighted and fused, that is, weights are assigned to each prediction task according to the degree of influence of each component on the overall equipment health condition. Usually, for components with higher risk or higher importance, the weights of their fault prediction results should be larger. By weighted-fusing the fault prediction results of each component, a comprehensive fault warning signal is calculated. This signal not only synthesizes the fault prediction results of each component but also takes into account the priorities and importance of different components.

[0067] Through the multi-fault prediction mechanism, the rotary drilling rig can independently predict the health status of each component and generate a comprehensive fault warning signal in combination with the weighted fusion method. This method not only improves the accuracy and reliability of the prediction but also provides timely and accurate fault warning information for the operator, helping to identify potential faults in advance, reduce equipment downtime, and ensure the safety and efficiency of the operation.

[0068] Furthermore, step P40 of the embodiment of the present application further includes: The multi-fault prediction mechanism relies on a multi-task learning model, which includes a fault task processing layer for multiple related prediction tasks and an independent parameter sharing layer. Among them, the multiple related prediction tasks share underlying parameters through the parameter sharing layer.

[0069] Specifically, the multi-fault prediction mechanism relies on a multi-task learning model, which processes multiple related fault prediction tasks by sharing underlying parameters. Through this architecture, multiple tasks can share underlying parameters, thereby improving the learning efficiency of the model and reducing the computational overhead.

[0070] In this multi-task learning model, first, an independent fault task processing layer is designed for each related fault prediction task. These processing layers are respectively responsible for processing the input data of each task and generating corresponding fault prediction results. For example: The hydraulic system fault task processing layer is used to receive data from the hydraulic system (such as pressure, flow rate, etc.) and predict potential faults in the hydraulic system, such as leakage, blockage, etc. The power system fault task processing layer is used to receive data from the engine or power system (such as load, speed, temperature, etc.) and predict whether there are faults such as overload and overheat in the engine. The mechanical component fault task processing layer is used to receive sensor data from mechanical components (such as acceleration, vibration, etc.) and judge whether there is structural damage or component wear. The control system fault task processing layer is used to receive data from the control system (such as electrical signals, system response time, etc.) and predict whether the control system will fail.

[0071] Each task processing layer extracts features and learns fault patterns from the input data through its specific network structure (such as convolutional neural network, long short-term memory network, etc.). Although each task has its own goals and characteristics, by sharing parameters, the model can better learn common features and improve the accuracy of fault prediction.

[0072] In the multi-task learning model, the parameter sharing layer is the part shared by all tasks and is responsible for extracting shared features from the underlying input data. Through this shared layer, multiple tasks can share the underlying shared representation, that is, information such as the overall health status of the device, the operating environment, and the dynamic changes of the device. For example, when dealing with the fault prediction of the hydraulic system of a rotary drilling rig, the operation path and environmental factors may affect multiple tasks, and this information is uniformly extracted in the parameter sharing layer to ensure information sharing between tasks.

[0073] In the parameter sharing layer, neural network layers can be used to process these common features. During the training process of the multi-task learning model, through the shared underlying network layer, the model will learn the correlations between these tasks. For example, the health status of the hydraulic system may be related to the wear degree of mechanical components, and the control system failure may be related to the load fluctuation of the power system. By sharing parameters, the model can transfer these implicit relationships between multiple tasks, improving learning efficiency and prediction accuracy.

[0074] Since multiple tasks share the underlying parameters, the multi-task learning model can effectively utilize the correlations between tasks. During the training process, multiple tasks will optimize the shared parameters in parallel, so that the error functions of different tasks are minimized simultaneously in the underlying network. This collaborative training can not only improve the accuracy of individual tasks but also help the model better learn the potential connections between tasks. For example, the fault of the hydraulic system may affect the operation of mechanical components, so the prediction task of the hydraulic system may provide additional information for the fault prediction task of mechanical components, and vice versa.

[0075] Through this underlying parameter sharing, the multi-task learning model can effectively reduce the consumption of computing resources while improving the accuracy of the prediction system. The shared parameters and features also enable the model to identify the common features between different tasks and optimize the learning process of the entire fault prediction system.

[0076] In summary, the embodiments of this application have at least the following technical effects: This application is based on a multi-source sensing module, which collects the hydraulic system, sensors, terrain, and historical operation and maintenance data of a rotary drilling rig in real time, preprocesses the data and extracts multiple features, including terrain influence, hydraulic system, and ground environment features. Then, based on the extracted features, dynamic modeling and health assessment are carried out to obtain the health status assessment result of the rotary drilling rig. Finally, according to the health status assessment result, a multi-source fault prediction mechanism is used to predict equipment faults and generate fault warning signals.

[0077] It achieves the technical effect of improving the real-time performance and accuracy of fault warning of a rotary drilling rig by fusing multi-source sensing data and an adaptive update mechanism for fault prediction and health management of the rotary drilling rig.

[0078] It should be noted that the above sequence of embodiments of this application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0079] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included within the protection scope of this application.

[0080] This specification and the drawings are only exemplary descriptions of this application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art can make various changes and modifications to this application without departing from the scope of this application. Thus, if these modifications and variations of this application fall within the scope of this application and its equivalent technologies, this application is intended to include these changes and modifications.

Claims

1. A method for predicting the faults of a rotary drilling rig under multi-source data analysis, characterized in that, The method includes: Based on a multi-source sensing module, multi-source sensing data of the target rotary drilling rig is collected in real time. The multi-source sensing data includes hydraulic system data, equipment sensing data, terrain sensing data, and historical operation and maintenance data; After preprocessing the multi-source sensing data, multi-source feature information is extracted. The multi-source feature information includes terrain influence features, hydraulic system features, and ground environment features; Based on the multi-source feature information, dynamic modeling and health assessment are performed to obtain a health status assessment result of the target rotary drilling rig; According to the health status assessment result, through a multi-source fault prediction mechanism, equipment fault prediction is performed to generate a fault warning signal.

2. The rotary drilling rig fault warning method under multi-source data analysis according to claim 1, characterized in that, After preprocessing the multi-source sensing data, multi-source feature information is extracted, including: Deploy multiple edge computing nodes in the control system of the target rotary drilling rig. The edge computing nodes are communicatively connected to the multi-source sensing module and can directly obtain data from the sensors; The edge computing nodes receive the multi-source sensing data and perform data preprocessing at each node to generate standardized multi-source data; On the edge computing nodes, multi-source feature information is extracted from the standardized multi-source data to obtain terrain influence features, hydraulic system features, and ground environment features, and the terrain influence features, hydraulic system features, and ground environment features are converted into a multi-source feature matrix.

3. The method for predicting the failure of a rotary drilling rig under multi-source data analysis according to claim 2, characterized in that, Based on the multi-source feature information, dynamic modeling and health assessment are performed to obtain a health status assessment result of the target rotary drilling rig, including: Based on the key components of the target rotary drilling rig, a multi-physics field model is established to establish a preliminary dynamics model; Multi-source feature information is extracted through the multi-source feature matrix to dynamically empower the preliminary dynamics model to generate a dynamic working model; According to the multi-source sensing data, the dynamic working model is used to simulate the operation path, and simulation data is collected for multi-level real-time health assessment to generate a health status assessment result of the target rotary drilling rig.

4. The method for predicting the failure of a rotary drilling rig under multi-source data analysis according to claim 3, characterized in that, Multi-source feature information is extracted through the multi-source feature matrix to dynamically empower the preliminary dynamics model to generate a dynamic working model, including: Based on the multi-source feature matrix, terrain influence features, hydraulic system features, and ground environment features are extracted; According to the terrain influence features, hydraulic system features, and ground environment features, variable factor quantization is performed to generate multiple model parameter adjustment coefficients. The model parameter adjustment coefficients include terrain adjustment coefficients, soil adjustment coefficients, load adjustment coefficients, and hydraulic adjustment coefficients; The multiple model parameter adjustment coefficients are used to dynamically empower the preliminary dynamics model to generate the dynamic working model.

5. The fault warning method for a rotary drilling rig under multi-source data analysis according to claim 3, wherein The dynamic working model is embedded with an adaptive update mechanism. Based on the adaptive update mechanism, continuously updated multi-source feature information is received in real time to perform incremental learning and dynamic update on the dynamic working model.

6. The method for predicting the failure of a rotary drilling rig under multi-source data analysis according to claim 3, wherein, According to the multi-source sensing data, the dynamic working model is used to simulate the operation path, and simulation data is collected for multi-level real-time health assessment, including: Based on the multi-source sensing data, the operation path of the target rotary drilling rig is planned to generate an optimal planned path; According to the optimal planned path, use the dynamic working model to simulate the operation path, and during the simulation process, collect multivariate component state data sets respectively according to the equipment component levels; Based on the multivariate component state data sets, conduct multi-level real-time health assessment on the target rotary drilling rig, including conducting power system health assessment, hydraulic system health assessment, control system health assessment, and mechanical component health assessment, and fuse the multi-level assessment results to generate the health state assessment result of the target rotary drilling rig.

7. The rotary drilling rig fault warning method under multi-source data analysis according to claim 6, characterized in that, According to the health state assessment result, through a multivariate fault prediction mechanism, conduct equipment fault prediction and generate a fault warning signal, including: According to the equipment component levels, generate multiple relevant prediction tasks and establish a multivariate fault prediction mechanism based on this; Through the multivariate fault prediction mechanism, simultaneously conduct fault prediction on the multiple relevant prediction tasks to generate multivariate fault prediction results; Weight and fuse the multivariate fault prediction results to generate the fault warning signal.

8. The rotary drilling rig fault warning method under multi-source data analysis according to claim 7, characterized in that, The multivariate fault prediction mechanism relies on a multi-task learning model, which includes fault task processing layers for multiple relevant prediction tasks and an independent parameter sharing layer. Among them, the multiple relevant prediction tasks share underlying parameters through the parameter sharing layer.

Images