Antarctic ice layer gas and microorganism detection method based on hot melt drilling digital twinning

Abstract

The invention discloses an Antarctic ice layer gas and microorganism detection method based on hot-melt drilling digital twinning, and belongs to the technical field of digital technologies and polar scientific investigation engineering. Parameters are collected in real time through various sensors integrated in a hot-melt drill bit and are transmitted to a far-end console through electromagnetic waves; and a complete information flow is presented after the processing of an anomaly checking algorithm and an AI decision early warning system. The system not only can simulate the influence of different temperatures, pressures and drilling speeds on an ice layer so as to optimize a drilling parameter combination in real time, but also can synchronously map key indexes such as drill bit wear and system efficiency in combination with Internet of Things sensor data and a digital twinning technology. Besides, the collected bubble samples are subjected to component analysis through an integrated sensor array, and the characteristics of the microflora are dynamically displayed by using a metabolite identification model, so that accurate data support is provided for polar scientific investigation.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of digital technology and polar scientific research engineering, and particularly relates to a method for detecting Antarctic ice layer gas and microorganisms based on hot melt drilling digital twinning. BACKGROUND

[0002] The deep Antarctic ice layer contains a large number of gas samples and rich microbial communities. The bubbles in the ice core perfectly preserve various components in the ancient atmosphere, and by analyzing the bubble samples, the ancient climate fluctuation law can be traced, and the types of ancient microbial communities under the ice layer can be inferred according to the content and concentration of different gases in the bubble samples.

[0003] The traditional mining and testing technology is based on ice core drilling technology, that is, the ice layer is drilled by using mechanical rotary drilling technology. After the bubbles in the ice core are collected, the bubbles are physically broken in a laboratory environment to release the gas, and the corresponding gas components are directly measured by using isotope analysis method. This method relies on manual experience judgment and is prone to repeated operation due to ice layer mutation. Meanwhile, the isolation of snow layer and the installation of casing require very fine operation, otherwise the external environment will be disturbed or the gas in the sample will be leaked.

[0004] The emergence of hot melt drill bit greatly reduces the above problems. In some scenarios, the ice layer is softened by using electric heating or steam, thereby improving the drilling efficiency, reducing the impact on the environment, and reducing the consumption of mining cost. However, during the working process of the drill bit, the conduction efficiency of the hot melt equipment is easily reduced due to the influence of the extremely low temperature in the Antarctic, and the low-temperature embrittlement of the metal parts is accelerated, thereby increasing the risk of mechanical failure. The sensor is also prone to internal circuit fracture due to thermal expansion and cold contraction fatigue caused by frequent cold and hot alternation, thereby causing unstable voltage at both ends of the sensor or damage to the sensor during detection, and affecting the accuracy of gas data collection during detection. Therefore, a method for real-time detection of the running state of each equipment part is needed to ensure that the risk of machine failure can be sensed in time, and the risk prevention and control can be prepared in advance to improve the reliability and precision of gas and microorganism detection.

[0005] The core of digital twinning technology is to integrate a physical model, acquire environmental information and part running information through a sensor, and realize real-time synchronous digital mapping technology in a virtual space with a physical entity. The various parameters of the drill bit are collected in real time by using various sensors at the end of the hot melt drill bit, the whole process information is presented on the remote console after transmission via an information channel, abnormality detection algorithm and intelligent decision and early warning processing, the whole collection and testing process is realized full-range and intelligent supervision and early warning, and the detection precision of the Antarctic deep ice layer gas and the key data such as the gene of the subsequent analysis of ancient microorganisms have higher precision and reliability. SUMMARY

[0006] The purpose of the present application is to provide a technical detection method that not only can solve the problem of the prior art that cannot timely feedback operation component information, but also can use abnormality checking algorithm and intelligent decision module to give early warning and optimization for the whole process of drilling, at the same time, use high-precision gas analysis sensor array integrated in hot melt drill bit to accurately analyze the collected bubble samples, and use metabolic product identification model to specifically analyze the types and contents of microorganisms.

[0007] The Antarctic ice layer gas and microorganism detection method based on hot melt drilling digital twinning of the present application comprises the following steps:

[0008] 1) Equipped with sensors suitable for detection requirements, including: low-temperature-resistant and anti-interference temperature sensors, pressure sensors, vibration sensors, used for real-time detection of various parameters of the drill bit in the mining process; ice layer analysis sensors, used for real-time feedback of related information of ice layer thickness and ice layer crack distribution; high-precision gas analysis sensors, which are used for detecting the gas composition in the deep ice bubble; equipped with redundant execution structure to cope with single point failure, to ensure that the overall work process is not affected when part of the sensors fail; solid-phase microextraction fiber is installed in this area to enrich the collected gas in the sample;

[0009] 2) Transmission of deep ice layer gas collection data, comprising the following steps:

[0010] 2.1 Use low-frequency electromagnetic wave communication to adjust the transmission frequency to match the ice layer medium characteristics, and realize penetration of several kilometers;

[0011] 2.2 Enable emergency communication when the main transmission fails, the emergency communication adopts a radar wave relay system, carries communication signals through pulse modulation technology, and establishes a relay link through ice layer-bedrock interface reflection;

[0012] 2.3 Data encryption and authority management: data transmission encryption selects a special encryption protocol, storage encryption selects static data encryption, and local database and cloud files are encrypted using an algorithm; the authority management adopts role-based access control, and assigns function authorities according to the organization structure in the whole investigation team;

[0013] 3) Management of data in the whole life cycle, through data collection, storage, processing and analysis links, the transmitted data are structured, comprising the following steps:

[0014] 3.1 The data collected in the data, for example, temperature, pressure sensor data need real-time feedback device state, using edge computing method for local processing, the integration of multi-source heterogeneous data into a unified format or summary information technology to reduce the transmission and operation load of cloud big data platform;

[0015] 3.2 The data backup after local processing, transmission to the cloud big data platform, using edge computing embedded exception detection algorithm, the backup data for abnormal detection, the purpose is to realize millisecond level response for the abnormal data in the device transmission parameters, avoid the decision lag caused by cloud transmission delay, that is, the standardization processing of multi-source data stream sensor time series data or log text, through statistical model, machine learning, deep learning to analyze and feedback the possible abnormal data and situation;

[0016] 3.3 Using metabolic product recognition model, the gas under deep ice layer is analyzed, and the characteristics of the colony under deep ice layer are inferred according to the proportion of the characteristic products in the gas;

[0017] 3.4 The analysis data and microbial community data of the gas in the ice layer are uploaded to the cloud big data platform at the same time, the abnormal values are eliminated, the data format is unified, the sorting, screening and calculation are used to optimize the data structure, and the statistical analysis, machine learning algorithm is used to arrange the data, and finally the processed results are presented on the remote console;

[0018] 4) Construct digital twin engine, including the following modules:

[0019] 4.1 Drill abnormal state detection module: using multi-physical field simulation model, that is, through numerical method to simulate physical field, reveal the coupling effect of complex system in cross scale, adopt heat-force coupling model, fit the ice layer melting and drill deformation; Combined with the material, stress direction and angle of hot melt drill bit, sensor output signal abnormal rate, to comprehensively judge the service life of hot melt drill bit; In the case of integrating temperature, pressure, vibration sensor in drill bit body, equipped with laser displacement sensor to measure hole displacement in real time, realize millisecond level capture of drill distance error; Deploy chip at the end of the device, denoise the original signal, extract the dynamic radius of hot softening area and axial thrust fluctuation frequency parameters, judge the resistance of material penetration and the stability of material contact; Summarize the output data of each sensor to set the safety operation threshold, and use the abnormal detection algorithm to detect the abnormal state of the drill parts;

[0020] 4.2 Drilling parameter real-time optimization module: through the core data collected by the performance sensor, the drilling tool working parameters obtained by the sensor network and the ice layer information, the correlation analysis method is used to determine the correlation matrix between temperature, pressure, drilling speed and ice layer, a multi-objective optimization model is used to set drilling speed, energy consumption and drill bit life as the optimization objective function, and the ice layer drillability coefficient is introduced to correct the model parameters, and finally the optimal combination of drilling tool parameters in the drilling process is obtained by using the optimization algorithm to process the constraint conditions;

[0021] 4.3 Multi-component gas precision analysis module: the high-precision gas analysis sensor integrated in the hot melt drill bit is used to analyze the gas samples collected from the deep ice layer; a real-time gas analysis platform is established, the analysis results of each sensor are summarized, and the data are sent to the cloud big data platform for subsequent analysis of microbial species and content;

[0022] 4.4 Microbial analysis module: based on the metabolic product identification model, the high-precision gas analysis sensor is used to detect the samples after gas concentration and separation; a real-time gas analysis platform is constructed, a dynamic threshold is set, and the microbial content and species contained are analyzed according to the identified gas sample results;

[0023] 4.5 Cloud data storage module: store the sensor device parameters processed by edge computing, the gas parameters detected by high-precision gas analysis sensor, and the microbial species and content parameters analyzed by metabolic product identification model, so that when the data in the collection process is abnormal, the original data can still be referred to;

[0024] 4.6 In the application layer, a visual interface AR / VR is assembled, the content is more intuitively presented by combining a multi-physical simulation model; the remote console can remotely adjust and urgently intervene the parameters in the system; after the data collected by the sensor are transmitted and processed, they are finally displayed on the data analysis instrument panel.

[0025] Compared with the traditional detection and collection method, the digital twin can not only detect the state of the drilling tool in real time through the performance sensor, map the drill bit wear, system efficiency and other key indicators, but also simulate the optimal scheme of the drilling parameter combination by establishing the correlation matrix and the optimization model to reduce the equipment downtime risk in extreme environment. At the same time, the built-in high-precision gas analysis sensor array can analyze the bubbles in the collected samples in real time, and the metabolic product identification model can dynamically display the characteristics of the microbial community. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 The flowchart of the Antarctic ice layer gas and microbial detection method based on the hot melt drilling digital twin;

[0027] Figure 2Flow chart for drilling tool anomaly detection and parameter optimization

[0028] Figure 3 Flow chart for anomaly detection algorithm embedded in edge computing

[0029] Figure 4 Flow chart for drilling parameter optimization algorithm

[0030] Figure 5 Flow chart for multi-component gas precision analysis

[0031] Figure 6 Flow chart for metabolic product identification model DETAILED DESCRIPTION

[0032] The present application is a hot melt drilling digital twin-based Antarctic ice layer gas and microorganism detection method. The application not only can collect and transmit the related parameters of the hot melt drill bit in real time, but also can detect and optimize the whole drilling process by combining the Internet of Things. At the same time, the application uses integrated high-precision gas analysis sensors to analyze the collected bubble samples, and uses metabolic product identification to obtain the dynamic characteristics of the microbial community. The following describes the core technology in the digital twin engine module of the application in conjunction with the accompanying drawings.

[0033] Anomaly detection algorithm implementation

[0034] ①Collect sensor time series data and log text from multiple data streams to ensure the comprehensiveness and accuracy of the data. Analyze the definition of abnormal data and the scene, first confirm the data anomaly type: single data point anomaly, such as sensor noise; specific context anomaly, such as sudden temperature drop and sharp rise; a complete set of data overall anomaly, such as network traffic flood attack;

[0035] ②Data preprocessing: fill in the missing values of the collected data to ensure data integrity; standardize different types of data to eliminate the influence of different dimensions; use sliding window to segment the data for subsequent analysis;

[0036] ③Data feature engineering stage, extract time domain features to capture the law of data change over time; apply frequency domain wavelet transform to extract frequency features of data;

[0037] ④After feature extraction, based on statistical model for rapid screening, time series analysis and threshold determination. Z-Score statistical model is a description of the position of a data relative to the average value of the entire data set, which represents the distance between the original data and the average value. In the anomaly detection algorithm, the formula and Calculate the mean and standard deviation of the data, and set a threshold for the statistical model. Data points that exceed the threshold range are marked as outliers.

[0038] ⑤ Use machine learning techniques to build complex models to identify dynamic characteristics. Isolation forest algorithm isolates abnormal points by randomly dividing feature space. A number of binary trees are established for randomly selected sub-samples in the data set. The average path length of each data point in all trees is used to calculate the anomaly score.

[0039] ⑥ Use deep learning algorithms to extract high-dimensional data features and integrate multi-modal data. Use auto-encoder for deep learning method to train the network to compress input data into low-dimensional code and reconstruct it. In addition, the error (MSE) of reconstruction needs to be calculated, and the points with large error are considered abnormal. LSTM time series detection uses historical sequences to predict future values. If the residual error between the actual value and the predicted value exceeds the threshold, it is considered abnormal.

[0040] ⑦ After cross-checking the data in three ways, dynamically adjust the model threshold.

[0041] ⑧ Model training and verification, learning without human supervision, such as using all data for training without any labels; still need to rely on manual review for verification; under supervised learning, need to divide training set and validation set, and use precision and recall as indicators; semi-supervised learning, only use normal data to train the model, and output the probability of abnormality during detection;

[0042] ⑨ Deploy and iterate the entire system, integrate it lightly on the basis of ensuring the original business capabilities, and realize the closed loop of detection-warning-disposal; continuous optimization mechanism, online learning updates model parameters, adapts to data distribution drift, regularly analyzes false positive cases, optimizes feature engineering and logic algorithms.

[0043] Real-time drilling parameter optimization algorithm:

[0044] ① Use temperature sensors, pressure sensors, vibration sensors and other performance sensor arrays deployed in hot melt drill bits to collect temperature gradient, pressure, and drilling speed in real time, and establish a standardized data warehouse for the collected data;

[0045] ② Use grey correlation analysis method to determine the correlation matrix between drilling pressure, drilling speed, temperature and ice layer, and construct a Pareto multi-objective optimization model. Set drilling speed, energy consumption, and drill bit life as optimization target parameters, and introduce ice layer drillability coefficient to correct model parameters;

[0046] ③ Use improved NSGA-II algorithm to handle constraints and generate non-inferior solution set; based on ideal reference point theory, screen the solution set to balance the priority of parameters;

[0047] ④ Dynamically update the weight coefficients of the optimization model through machine learning to adapt to the change in the depth of the hot melt drill bit under the ice;

[0048] ⑤ Automatically adjust the system output according to the optimization results.

[0049] Multi-component gas precision analysis module:

[0050] ① Use the anti-pollution hot melt drilling technology to directly obtain the gas and its water samples from the deep ice layer in the Antarctic without being contaminated by surface microorganisms;

[0051] ② Concentration and separation of gas, use ultra-low temperature pre-concentration device to capture trace gas, improve the detection sensitivity of low concentration metabolites;

[0052] ③ Use infrared absorption sensor to detect gas, use the absorption characteristics of gas molecules to specific wavelength infrared light, measure the degree of light intensity attenuation to quantitatively analyze gas composition, such as through detecting the change of gas temperature or pressure to determine the content, for carbon monoxide, ethanol and hydrocarbon gas;

[0053] ④ Use electrochemical sensor to test gas, based on the redox reaction of gas in electrolyte solution, quantify gas concentration through electrode current change, for oxygen, carbon monoxide, hydrogen sulfide gas detection, precision can reach ppm level;

[0054] ⑤ Use semiconductor sensor to test gas, use the resistance change of gas sensitive material to respond to the adsorption of gas, for the detection of combustible gas (methane, hydrogen) and carbon monoxide;

[0055] ⑥ Calibrate and control the error of the test results, that is, dynamically set the threshold, according to the gas concentration range, grade the sensitivity of the sensor, establish a real-time gas analysis platform, and finally obtain the type and content of gas in the sample bubble.

[0056] Metabolite identification model:

[0057] ① Use high-precision gas analysis sensor to analyze the gas composition of the sample collected under the ice, and additionally install TDLAS laser sensor to verify the double gas concentration ratio of methane and carbon dioxide on the basis of multi-component gas precision analysis module;

[0058] ② Establish a microorganism concentration-gas feature library based on the data obtained by the gas analysis sensor;

[0059] ③ Predict the total number of microorganisms through LSTM neural network;

[0060] ④ Use random forest or XGBoost to predict the abundance of microorganisms and visualize the gas metabolism characteristics of different bacterial groups through PCA, t-SNE;

[0061] ⑤ Upload the metabolic network graph of the generated microbial metabolic gases-total number-abundance to the cloud big data platform.

Claims

1. A method for detecting gases and microorganisms in Antarctic ice based on hot melt drilling digital twins, characterized in that: The following steps are involved: 1) Equipped with sensors suitable for detection requirements, including: low-temperature and anti-interference temperature sensors, pressure sensors, vibration sensors, ice analysis sensors, and high-precision gas analysis sensors; equipped with redundant execution structures to deal with single point failures, ensuring that the overall work process is not affected when some sensors fail; solid-phase microextraction fibers are installed in this area to enrich the gas in the collected samples; 2) Transmitting the deep ice layer gas collection data, including the following steps: 2.1 Using low-frequency electromagnetic wave communication, the transmission frequency is adjusted to match the characteristics of the ice layer to achieve penetration of several kilometers; 2.2 Emergency communication is enabled when the main transmission fails. Emergency communication uses a radar wave relay system, which carries communication signals through pulse modulation technology and establishes a relay link through reflection at the ice-bedrock interface; 2.3 Data Encryption and Rights Management: Data transmission encryption uses a proprietary encryption protocol, while storage encryption uses static data encryption. Algorithms are used to encrypt local databases and cloud files. Rights management uses role-based access control, and functional permissions are allocated according to the organizational structure of the entire investigation team. 3) Manage the data throughout its life cycle, structuring the transmitted data through data collection, storage, processing, and analysis, including the following steps: 3.1 Use edge computing to locally process the collected data with high real-time requirements, and integrate multi-source heterogeneous data into a unified format or summary information technology; 3.2 After backing up the locally processed data, it is transferred to the cloud big data platform and anomaly detection is performed on the backup data using the nested anomaly detection algorithm in edge computing; 3.3 Using the metabolite identification model, analyze the composition of the gas under the deep ice layer and infer the characteristics of the bacterial colonies under the deep ice layer based on the proportion of characteristic products in the gas; 3.4 Upload the gas analysis data and microbial community data in the ice layer to the cloud big data platform for data cleaning or storage, and remove outliers. Standardize the data format and optimize the data structure through sorting, screening, and calculation methods. Analyze the data using statistical analysis and machine learning algorithms, and finally present the processed results on the remote console. 4) Build a digital twin engine, including the following modules: 4.1 Drill Tool Abnormal State Inspection Module: Utilizing a multi-physics simulation model and a thermal-mechanical coupling model, the module fits ice melting and borehole deformation. The module comprehensively determines the service life of the melt drill bit based on its material, force direction and angle, and sensor output signal abnormality rate. While the drill bit integrates temperature, pressure, and vibration sensors, it is equipped with a laser displacement sensor to measure hole offset in real time, enabling millisecond-level capture of drilling distance errors. Chips are deployed on the device side to perform noise reduction on the raw signal, extracting the dynamic radius of the thermal softening zone and the axial thrust fluctuation frequency parameters to determine the resistance to material penetration and the stability of contact with the material. The output data of each sensor is aggregated to set safe operation thresholds, and an abnormality inspection algorithm is used to inspect drill tool components for abnormal conditions. 4.2 Real-time Drilling Parameter Optimization Module: Using core data collected by performance sensors, drill tool operating parameters acquired by the sensor network, and ice layer information, the module uses correlation analysis to determine the correlation matrix between temperature, pressure, drilling rate, and ice layer. A multi-objective optimization model is then used to set drilling rate, energy consumption, and drill bit life as optimization objective functions. The ice layer drillability coefficient is then introduced to modify the model parameters. Finally, an optimization algorithm is used to address the constraints to determine the optimal combination of drilling tool parameters during the drilling process. 4.3 Multi-component gas precision analysis module: Utilizes the high-precision gas analysis sensor integrated in the melt drill bit to analyze gas samples collected from deep ice layers; establishes a real-time gas analysis platform to aggregate the analysis results of each sensor, organize the data, and send it to the cloud big data platform for subsequent analysis of microbial species and content; 4.4 Microbial Analysis Module: Based on the metabolite identification model, high-precision gas analysis sensors are used to detect samples that have undergone gas concentration and separation. A gas analysis platform is built in real time, dynamic thresholds are set, and the content and types of microorganisms contained in the identified gas samples are analyzed based on the results. 4.5 Cloud data storage module: stores sensor equipment parameters processed by edge computing, gas parameters detected by high-precision gas analysis sensors, and microbial species and content parameters analyzed using metabolite identification models, ensuring that if any abnormalities occur in the data during the collection process, the original data is still available for reference; 4.6 The application layer incorporates AR / VR visualization interfaces, combined with multi-physics simulation models to present content more intuitively. The remote control console can remotely adjust system parameters and conduct emergency interventions. The data collected by sensors is transmitted and processed, and ultimately displayed on the data analysis dashboard.

Images