A method, system and device for regulating a while-drilling helium collection device, and a storage medium

Abstract

The present disclosure relates to the field of intelligent regulation of a while-drilling helium collection device, and discloses a while-drilling helium collection device regulation method, system, device and storage medium. The method comprises: acquiring monitoring data of the while-drilling helium collection device; constructing a characteristic quantity reflecting fluid characteristics based on the monitoring data; training a collection device control parameter prediction model based on the relationship between the monitoring data and the characteristic quantity and the collection device control parameter; predicting a target control parameter based on the prediction model; and dynamically regulating the while-drilling helium collection device based on the target control parameter. The present disclosure automatically adjusts ultrasonic parameters based on real-time monitoring of while-drilling helium collection device operating parameters, thereby avoiding energy waste or insufficient efficiency caused by fixed parameters.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent control technology for helium collection devices while drilling, specifically to a control method, system, equipment, and storage medium for helium collection devices while drilling. Background Technology

[0002] As the lightest inert gas in nature, helium plays an irreplaceable role due to its unique physicochemical properties. In the field of low-temperature superconductivity, liquid helium is an essential medium for maintaining the operation of superconducting magnets, and 75% of the world's helium is used in high-end medical and scientific research devices such as nuclear magnetic resonance imaging (NMR) equipment and particle accelerators. In semiconductor manufacturing, high-purity helium serves as a protective gas to ensure the stability of wafers during high-temperature chemical vapor deposition. In the aerospace field, helium's low-density properties are relied upon for applications in satellite cooling systems and rocket fuel pressurization.

[0003] Currently, some patented technologies have proposed ultrasonic-based measurement-while-drilling (MWD) or fluid handling solutions, but they still have many limitations in achieving precise and adaptive control. Existing technical solutions can be mainly divided into the following categories:

[0004] Patent IN3193DEL2013A discloses a bulk flow reactor capable of continuously processing large volumes of fluid-mediated reactants via ultrasonic cavitation. This process requires close mixing or activation of the reactants and includes, but is not limited to, chemical or biochemical reactions, leaching, small particle cleaning, wastewater treatment, sludge separation, particle decomposition, nanoparticle preparation, colloid formation, emulsification, and contaminated liquid treatment, all in an energy-efficient manner. The patent discloses a three-chamber design with an innovative arrangement of parallel and densely packed ultrasonic processing modules for the isolated and efficient injection of ultrasonic energy into the medium. This patent provides a facility for expanding processing capacity without efficiency loss due to increased power of individual transducers. A further innovation of this patent is the provision of separate temperature profiles in three zones for laboratory, pilot plant, and industrial processes requiring such control.

[0005] US Patent 4455268A discloses a method for curing a structure formed of fiber-reinforced composite materials or the like in an autoclave, comprising the following steps: measuring the ambient temperature in the autoclave; measuring the ambient pressure in the autoclave; measuring the temperature on the surface of the structure being cured; measuring the time elapsed during the curing process of the structure; comparing the measured ambient temperature with an ambient temperature value selected from a predetermined ambient temperature versus time curve for the measured elapsed time at least periodically; comparing the measured surface temperature value of the structure with a surface temperature value selected from a predetermined surface temperature versus time curve for the measured elapsed time at least periodically; adjusting the heat supplied to the autoclave so that the measured ambient temperature and the surface temperature of the structure are within a predetermined range of variation of the selected values; determining the instantaneous viscosity of the structure at least periodically; comparing the determined viscosity with a reference viscosity at least periodically; setting a reference pressure value based on the comparison result between the determined viscosity and the reference viscosity; comparing the ambient pressure in the autoclave with the reference pressure value; and adjusting the ambient pressure in the autoclave so that the measured ambient pressure is within a predetermined range of variation of the reference pressure value. This patent allows for the control of autoclave input parameters (i.e., pressure and temperature) based on reliable and accurate monitoring, rather than simply prediction, of desired output parameters (i.e., viscosity). Furthermore, by real-time monitoring of the viscoelasticity of the composite material and controlling the autoclave pressure in real-time based on the results of this monitoring, optimal pressure conditions can be reliably obtained to minimize the formation of micropores during the curing process. The patent's dual pressure control mechanism allows for pressure control based on empirically derived pressure curves and specified viscosity behavior curves, thereby improving the accuracy and reliability of the pressure control mechanism. In addition, the ability to measure component temperature and autoclave ambient temperature provides additional flexibility for interactive operator control based on maximum temperature overshoot and component surface temperature differences.

[0006] However, existing technologies still have the following problems:

[0007] The existing technologies suffer from several shortcomings: First, they lack a single control dimension and multi-parameter fusion. They fail to comprehensively consider the synergistic and interactive effects of multiple key parameters such as viscosity, particle concentration, temperature, and pressure, resulting in one-sided control decisions and an inability to adapt to complex helium extraction processes. Second, their models lack expressive power. Relying on fixed formulas or simple linear models, they cannot effectively learn and handle the complex nonlinear relationships between fluid properties and ultrasonic parameters, leading to poor prediction accuracy and adaptability. Third, they lack uncertainty management mechanisms. Existing technologies treat model predictions as absolutely reliable instructions, ignoring the inherent uncertainty in complex environments. When uncertainty is high, blindly following predictions can lead to system oscillations or equipment damage, lacking robust fault-tolerant control strategies. Fourth, their adjustment strategies are rigid and lack intelligence. The existing adjustment process fails to combine predicted values ​​with the confidence level (uncertainty) of the prediction results, failing to intelligently balance aggressive adjustments with conservative stabilization, resulting in insufficient overall adaptability and stability. Summary of the Invention

[0008] This disclosure provides a method, system, device, and storage medium for controlling a helium collection device while drilling, in order to solve or alleviate one or more of the above-mentioned technical problems in the prior art.

[0009] According to one aspect of this disclosure, a method for controlling a drilling helium collection device is provided, comprising:

[0010] Acquire monitoring data from the drilling helium acquisition system;

[0011] Based on the monitoring data, characteristic quantities reflecting fluid properties are constructed;

[0012] Based on the relationship between the monitoring data and feature quantities and the control parameters of the acquisition device, a prediction model for the control parameters of the acquisition device is trained.

[0013] Predict the target control parameters based on the prediction model;

[0014] Dynamically regulate the helium collection device while drilling based on target control parameters.

[0015] In one possible implementation, the monitoring data includes the viscosity, temperature, particle concentration, and pressure of the fluid from which helium is to be extracted. Based on the monitoring data, characteristic quantities reflecting the fluid properties are constructed, including:

[0016] A viscosity-temperature influence factor is constructed based on the viscosity and temperature of the fluid to be helium extracted;

[0017] The particle pressure resistance factor is constructed based on the fluid particle concentration and pressure.

[0018] In one possible implementation, based on the relationship between the monitoring data and feature quantities and the control parameters of the acquisition device, a prediction model for the control parameters of the acquisition device is trained, including:

[0019] A neural network model is established, which includes an input layer, multiple hidden layers, and an output layer;

[0020] The monitoring data and features are used as input data for the input layer of the neural network, and the control parameters are used as output data for the output layer. The weights of each connection in the neural network model are continuously adjusted based on gradient descent and the Adam optimizer to minimize the loss function of the neural network model and obtain the prediction model.

[0021] In one possible implementation, predicting the target control parameters based on the prediction model includes:

[0022] By using Monte Carlo Dropout technology, multiple forward propagations are performed during the inference phase of the prediction model, and some network nodes are discarded to obtain the mean and standard deviation of the predicted values ​​of multiple control parameters.

[0023] Based on the mean and standard deviation of the predicted values, and combined with the preset adjustment strategy and uncertainty tolerance, the target control parameters are dynamically calculated.

[0024] According to one aspect of this disclosure, a control system for a drilling helium gas acquisition device is provided, comprising: an acquisition unit for acquiring monitoring data of the drilling helium gas acquisition device;

[0025] The construction unit is used to construct characteristic quantities reflecting fluid properties based on the monitoring data;

[0026] The training unit is used to train a prediction model for the control parameters of the acquisition device based on the monitoring data and the relationship between the feature quantities and the control parameters of the acquisition device.

[0027] The prediction unit is used to predict target control parameters based on the prediction model.

[0028] The control unit is used to dynamically control the helium collection device while drilling based on target control parameters.

[0029] In one possible implementation, the monitoring data includes the viscosity, temperature, particle concentration, and pressure of the fluid from which helium extraction is to be performed, and the building unit includes:

[0030] The first construction module is used to construct a viscosity-temperature influence factor based on the viscosity and temperature of the fluid from which helium is to be extracted;

[0031] The second construction module is used to construct a particle pressure resistance factor based on the fluid particle concentration and pressure.

[0032] In one possible implementation, the training unit includes:

[0033] A module is provided for building a neural network model, which includes an input layer, multiple hidden layers, and an output layer.

[0034] The adjustment module is used to take the monitoring data and feature quantities as input data of the neural network input layer and the control parameters as output data of the output layer. Based on the gradient descent method and the Adam optimizer, it continuously adjusts the weights of each connection in the neural network model to minimize the loss function of the neural network model and obtain the prediction model.

[0035] In one possible implementation, the prediction unit includes:

[0036] The acquisition module is used to utilize Monte Carlo Dropout technology to perform multiple forward propagations and discard some network nodes during the inference phase of the prediction model, thereby obtaining the mean and standard deviation of the predicted values ​​of multiple control parameters.

[0037] The calculation module is used to dynamically calculate the target control parameters based on the mean and standard deviation of the predicted values, combined with a preset adjustment strategy and uncertainty tolerance.

[0038] According to one aspect of this disclosure, an apparatus is provided, comprising:

[0039] Processor and memory;

[0040] The memory is used to store computer programs, and the processor calls the computer programs stored in the memory to execute the drilling helium collection device control method described in any of the above-mentioned embodiments.

[0041] According to one aspect of this disclosure, a computer-readable storage medium is provided, wherein a computer program is stored therein, which, when executed by a processor, enables the processor to perform the drilling helium collection device control method described in any of the preceding claims.

[0042] This disclosure has the following beneficial effects: Based on a real-time monitoring intelligent control system (viscosity, particle concentration, pressure feedback), this disclosure automatically adjusts ultrasonic parameters (frequency, power density) to avoid energy waste or insufficient efficiency caused by fixed parameters; it introduces Monte Carlo Dropout technology to evaluate the uncertainty of model output, ensuring that a conservative adjustment strategy is adopted when the uncertainty is high, and combines it with a PID controller to dynamically adjust the frequency and power density, which not only is based on the predicted value, but also comprehensively considers the uncertainty of the model prediction, thus achieving a more stable and smooth system control.

[0043] Details of one or more embodiments of this application are set forth in the following drawings and description. Other features and advantages of this application will become apparent from the accompanying drawings. It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not intended to limit this disclosure. Attached Figure Description

[0044] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0045] Figure 1 This is a flowchart of a method for controlling a helium harvesting device while drilling, as described in this exemplary embodiment.

[0046] Figure 2 This is a schematic diagram of the drilling fluid helium ultrasonic extraction device according to this exemplary embodiment;

[0047] Figure 3 This is a schematic diagram of the ultrasonic control steps of this exemplary embodiment;

[0048] Figure 4 This is a feature impact analysis diagram of this exemplary embodiment;

[0049] Figure 5 This is a block diagram of a control system for a drilling helium gas acquisition device according to an exemplary embodiment of the present invention;

[0050] Figure 6 This is a schematic diagram of the structure of a device according to an exemplary embodiment of this invention. Detailed Implementation

[0051] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this disclosure more comprehensive and complete, and to fully convey the concept of the example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, apparatus, steps, etc., can be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0052] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware units or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0053] The flowchart shown in the attached diagram is merely an illustrative example and does not necessarily include all steps. For example, some steps may be broken down, while others may be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0054] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein.

[0055] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or device that includes a series of steps or sub-modules is not necessarily limited to those steps or sub-modules that are explicitly listed, but may include other steps or sub-modules that are not explicitly listed or that are inherent to such process, method, product, or device.

[0056] This disclosure provides a method for controlling a helium sampling device while drilling, which is exemplarily used to control the following ultrasonic-based helium sampling devices while drilling (such as...). Figure 2 (as shown)

[0057] The ultrasonic-based helium harvesting device for drilling includes a pre-separation unit and an ultrasonic degassing module. After the drilling fluid is drawn from the drill bit, it enters the pre-separation unit, where coarse particles in the drilling fluid are filtered out to prevent damage to the pipeline system and subsequent reaction chambers. The mud fluid after passing through the pre-separation unit enters the ultrasonic degassing module through the pipeline system, where helium is separated.

[0058] The pre-separation device includes a hydrocyclone separator and a buffer settling tank; the hydrocyclone separator uses a centrifugal force gradient (200-800G) to separate coarse solid particles (particle size >50μm), while the buffer settling tank further separates incompletely desorbed solid particles;

[0059] The ultrasonic degassing module includes a helium separation chamber, a helium storage chamber, an ultrasonic coupling plate, and an ultrasonic transducer array (frequency range 28-40kHz).

[0060] The helium separation chamber is connected to the helium storage chamber, and a solenoid valve for controlling the opening and closing of the connection channel is provided on the connection channel between the helium separation chamber and the helium storage chamber; the ultrasonic transducer array is disposed in the helium separation chamber, and the ultrasonic transducer array and the mud in the helium separation chamber are isolated by the ultrasonic coupling plate, so that the ultrasonic waves of the ultrasonic transducer array are conducted to the helium separation chamber through the ultrasonic coupling plate with an acoustic impedance between the transducer surface and the mud, thus avoiding direct contact between the ultrasonic transducer array and the mud;

[0061] The helium storage chamber is provided with a helium outlet, and a solenoid valve is provided at the helium outlet so that when the pressure in the helium storage chamber is greater than a predetermined value, the helium in the helium storage chamber is drawn out through the solenoid valve; correspondingly, a pressure sensor is provided in the helium storage chamber to detect the pressure inside the chamber.

[0062] The helium separation chamber is equipped with a liquid level sensor, a viscosity sensor, a particle concentration sensor, and a pressure sensor. The pressure sensor detects whether the pressure in the helium separation chamber exceeds a threshold. If it does, the solenoid valve is opened to discharge the helium from the separation chamber into the helium storage chamber. The viscosity sensor and particle concentration sensor monitor the viscosity and particulate matter in the slurry to provide feedback on the frequency and voltage of the ultrasonic transducer. The liquid level sensor monitors whether the liquid level in the helium separation chamber reaches a predetermined threshold. If it does, the injection of slurry into the helium separation chamber must be stopped. After separation is completed in the helium separation chamber, the slurry is discharged through the outlet of the helium separation chamber before slurry is injected into the helium separation chamber again.

[0063] The working process of the above device is as follows:

[0064] After the drilling fluid is drawn from the drill bit, it enters the pre-separation unit, where coarse particles are screened out to prevent damage to the pipeline system and subsequent cavities. The mud, after being processed by the pre-separation unit, enters the ultrasonic degassing module via the management system. First, the inlet solenoid valve of the helium storage chamber is closed, and the ultrasonic drive module is activated for degassing. During degassing, data from the level sensor, viscosity sensor, particle concentration sensor, and pressure sensor are monitored in real time, and the ultrasonic frequency and power density are dynamically adjusted accordingly to optimize its operation. This prevents the module from burning out due to excessive input power and avoids insufficient degassing efficiency due to insufficient input power. When the accumulated helium reaches the opening threshold of the helium storage chamber solenoid valve, the valve opens, allowing helium to enter the helium storage chamber for storage. When the pressure in the helium storage chamber reaches a critical value, the helium is discharged from the system.

[0065] In summary, this disclosure utilizes the matching relationship between acoustic characteristics and fluid dynamics to dynamically adjust the ultrasonic frequency and power density (15-25) by real-time monitoring of fluid conditions (such as changes in viscosity and particle concentration). This ensures that the cavitation intensity is always maintained within the optimal range. This technology avoids the problem of low energy utilization efficiency under fixed parameters and significantly improves the adaptability of the treatment process.

[0066] Secondly, by organically connecting the acoustic field zone and the mechanical separation zone, this technology enables efficient helium capture. In the ultrasonic pretreatment stage, the gas-solid-liquid mixture undergoes initial desorption and particle breakage; subsequently, the separation process is enhanced by centrifugal force gradient, utilizing the difference in fluid inertia to achieve three-phase stratification. This multi-stage synergistic mechanism significantly reduces system complexity while avoiding the risk of secondary pollution. Finally, erosion-resistant acoustic interface materials can be prepared using surface functionalization modification technology. By controlling the hydrophobicity, hydrophilicity, and microstructure of the material surface, the adhesion of mud particles to the separation interface can be significantly reduced.

[0067] Figure 1 This is a flowchart of a control method for a drilling helium gas acquisition device according to an exemplary embodiment of this invention, such as... Figure 1 As shown, an exemplary embodiment of this disclosure provides a method for controlling a helium harvesting device while drilling, comprising:

[0068] Acquire monitoring data from the drilling helium acquisition system;

[0069] Based on the monitoring data, characteristic quantities reflecting fluid properties are constructed;

[0070] Based on the relationship between the monitoring data and feature quantities and the control parameters of the acquisition device, a prediction model for the control parameters of the acquisition device is trained.

[0071] Predict the target control parameters based on the prediction model;

[0072] Dynamically regulate the helium collection device while drilling based on target control parameters.

[0073] This embodiment proposes a drilling helium acquisition method based on ultrasonic technology. In this method, the ultrasonic frequency and power density are dynamically adjusted by monitoring the viscosity, particle concentration, temperature and pressure in the helium separation chamber. It is worth noting that a temperature sensor is added to the drilling helium acquisition device to obtain temperature data.

[0074] like Figure 3 As shown, the specific steps are as follows:

[0075] Data acquisition: Viscosity sensors, particle concentration sensors, pressure sensors, and temperature sensors acquire data in real time and transmit the acquired data to the control system;

[0076] Data preprocessing: The collected data is preprocessed to remove noise and outliers, making the data more accurate;

[0077] The preprocessing includes using filtering techniques to remove noise and outliers from the collected data. The filtering uses a moving average filter to average the data over a window, reducing short-term fluctuations. Then, the filtered data is normalized.

[0078] Feature construction: New features are constructed by combining the relationship between viscosity and temperature, particle concentration and pressure to reflect the effects on ultrasonic frequency and power density.

[0079] This embodiment successfully solves a series of key problems in existing ultrasonic helium gas acquisition technology during drilling, such as single control dimension, poor model adaptability, lack of robustness, and poor response performance, by organically combining multi-sensor information fusion, neural network intelligent decision-making, uncertainty quantification assessment and classical control theory. Ultimately, it achieves a comprehensive improvement in helium gas acquisition efficiency, system adaptability and operational stability.

[0080] Specifically, the monitoring data includes the viscosity, temperature, particle concentration, and pressure of the fluid to be helium extracted. Based on the monitoring data, characteristic quantities reflecting the fluid properties are constructed, including:

[0081] A viscosity-temperature influence factor is constructed based on the viscosity and temperature of the fluid to be helium extracted;

[0082] The particle pressure resistance factor is constructed based on the fluid particle concentration and pressure.

[0083] like Figure 4 As shown, viscosity exhibits a non-linear relationship with temperature, especially at higher liquid viscosities, where temperature changes can significantly impact fluid flowability. Therefore, this relationship is captured through the interaction characteristics of viscosity and temperature.

[0084] ;

[0085] Where: VTI viscosity-temperature influence factor; The value represents viscosity, expressed in Pa·s; T represents temperature, expressed in K. Indicates the effect of temperature T on fluid viscosity The decay factor; the effect of temperature on viscosity is expressed through exponential decay;

[0086] The interaction between particle concentration and pressure: Under the combined influence of particle concentration and pressure, particle behavior (such as aggregation, collision, and dispersion) affects the efficiency of ultrasonic wave propagation; especially under high pressure, fluids with high particle concentrations exhibit greater impedance to ultrasonic wave propagation, thus affecting cavitation effects.

[0087] ;

[0088] In the formula: PPI represents the particle pressure resistance factor; P represents particle concentration; P represents pressure; where PPI reflects the influence of particle concentration and pressure on fluid behavior, especially the influence of particle concentration on ultrasonic cavitation effect under high pressure conditions; constructing this interaction feature helps the model capture the influence of particles on fluid impedance, and thus affect the modulation of ultrasound.

[0089] This embodiment comprehensively monitors four key parameters of the fluid in real time: viscosity, particle concentration, temperature, and pressure. It innovatively constructs viscosity-temperature interaction characteristics and particle concentration-pressure interaction characteristics, deeply integrating the coupling effects of multiphysics fields. Subsequently, leveraging the powerful nonlinear mapping capabilities of a neural network model, it accurately predicts the optimal ultrasonic frequency and power density under the current operating conditions. This method can comprehensively and accurately respond to the complex and ever-changing downhole environment, significantly improving ultrasonic cavitation effects and helium collection efficiency.

[0090] Specifically, based on the relationship between the monitoring data and feature quantities and the control parameters of the acquisition device, a prediction model for the control parameters of the acquisition device is trained, including:

[0091] A neural network model is established, which includes an input layer, multiple hidden layers, and an output layer;

[0092] The monitoring data and features are used as input data for the input layer of the neural network, and the control parameters are used as output data for the output layer. The weights of each connection in the neural network model are continuously adjusted based on gradient descent and the Adam optimizer to minimize the loss function of the neural network model and obtain the prediction model.

[0093] This embodiment specifically includes the following steps:

[0094] Model prediction: Based on the preprocessed data and the established correlation model between fluid properties (viscosity, particle concentration, temperature, pressure and interaction characteristics) and ultrasonic frequency and power density, the ultrasonic frequency and power density are predicted.

[0095] The association model adopts a neural network model, which, for example, includes an input layer, multiple hidden layers, and an output layer.

[0096] The input data for the input layer includes viscosity, particle concentration, temperature, pressure, the interaction characteristics of viscosity and temperature (i.e., viscosity-temperature influence factor), and the interaction characteristics of particle concentration and pressure (i.e., particle pressure resistance factor). Based on this, an input data vector is constructed as follows: ;in Indicates viscosity; VTI represents particle concentration; T represents temperature; P represents pressure; VTI represents viscosity-temperature effect factor; PPI represents particle pressure resistance factor.

[0097] There are three hidden layers, each followed by a ReLU activation function; each hidden layer contains multiple neurons.

[0098] The output layer consists of two neurons, which predict two targets: ultrasound frequency and ultrasound power density, respectively.

[0099] The model uses mean squared error as its loss function. During training, the weights of each connection in the neural network are continuously adjusted based on gradient descent and the Adam optimizer to minimize the loss function. That is, the gradient of each parameter of the network is calculated according to the loss function, and the parameters are updated so that the loss function gradually decreases. Through multiple rounds of iterative training, the network parameters are continuously optimized until the network converges, that is, the loss function reaches its minimum value.

[0100] By using a neural network model, the complex linear and nonlinear relationships in the input data are effectively processed, and the ultrasonic frequency and power density are accurately predicted. Based on the input data (viscosity, particle concentration, temperature, pressure and their interaction characteristics), the weights and biases obtained through training are dynamically adjusted, thus realizing the precise control of ultrasound in the helium gas collection process.

[0101] Dynamic adjustment: The ultrasonic frequency and power density of the ultrasonic transducer array are adjusted based on the predicted ultrasonic frequency and power density.

[0102] The neural network model used in this embodiment, trained with a large amount of data, is able to learn and capture the complex nonlinear relationship between fluid properties and ultrasonic parameters. Regardless of increases in fluid viscosity, sudden changes in particle concentration, or drastic changes in temperature and pressure, the system can calculate and output optimal control parameters in real time, demonstrating strong environmental adaptability. This data-driven intelligent regulation method overcomes the shortcomings of traditional fixed-formula or PID controllers in terms of poor adaptability to nonlinear and time-varying systems, ensuring the stability and reliability (robustness) of the system under various extreme conditions.

[0103] Specifically, predicting target control parameters based on the prediction model includes:

[0104] By using Monte Carlo Dropout technology, multiple forward propagations are performed during the inference phase of the prediction model, and some network nodes are discarded to obtain the mean and standard deviation of the predicted values ​​of multiple control parameters.

[0105] Based on the mean and standard deviation of the predicted values, and combined with the preset adjustment strategy and uncertainty tolerance, the target control parameters are dynamically calculated.

[0106] To address the uncertainty of the model output, this embodiment employs Monte Carlo Dropout to obtain the execution range of the output. Monte Carlo Dropout discards some nodes of the network during the inference phase to generate multiple prediction values, thereby calculating the uncertainty of the prediction. The output can be represented as:

[0107] ;

[0108] ;

[0109] In the formula: and These represent the mean values ​​of the predicted frequency and power density, respectively. and These represent the standard deviations of the predicted values ​​for frequency and power density, respectively. This represents the frequency prediction value; This represents the predicted power density value;

[0110] The predicted values ​​of frequency and power density will be used as target values ​​and will be achieved by adjusting the operating parameters of the ultrasonic transducer array.

[0111] The ultrasonic frequency adjustment strategy is as follows: The ultrasonic frequency is affected by fluid characteristics (such as viscosity and particle concentration). In particular, high-viscosity fluids require lower frequencies to improve the cavitation effect. The frequency adjustment of the transducer array is achieved through the following proportional relationship:

[0112] ;

[0113] In the formula: This indicates the adjusted ultrasonic frequency; Indicates the predicted frequency; This represents an empirical adjustment coefficient (e.g., considering the effects of temperature changes). The standard deviation of the frequency predictions reflects the uncertainty of the model predictions; This represents the mean of the frequency prediction values;

[0114] Ultrasonic power density adjustment strategy: The adjustment of ultrasonic power density needs to consider the fluid impedance and the cavitation effect of bubbles, especially under high particle concentration and high pressure conditions, where a higher power density is required. The power density adjustment formula is as follows:

[0115] ;

[0116] In the formula: This represents the actual adjusted power density; This represents the predicted power density; The adjustment factor representing power density; This represents the standard deviation of the power density prediction. This represents the mean of the predicted power density.

[0117] The uncertainty tolerance setting is as follows: when the uncertainty of the prediction result is large (i.e., or If the threshold is exceeded, a conservative approach is taken to adjust the transducer, such as adjusting only a small percentage, to avoid negative effects from over-adjustment.

[0118] For example, if the uncertainty is higher than 10%, the adjustment range can be limited to 5%-10% of the original forecast value to ensure stable system operation.

[0119] Based on the predicted adjustment values ​​of ultrasonic frequency and power density, a PID controller is used to control the frequency and power density of the ultrasonic transducer array to ensure smooth adjustment of the transducer frequency and power density and avoid excessive oscillation. Through the PID controller, the system can dynamically adjust the frequency and power density to quickly and stably reach the target values.

[0120] Therefore, this invention introduces Monte Carlo Dropout technology to assess the uncertainty of the model output, ensuring that a conservative adjustment strategy is adopted when the uncertainty is high. It also combines a PID controller to dynamically adjust the frequency and power density, taking into account not only the predicted value but also the uncertainty of the model prediction, thus achieving a more stable and smooth system control.

[0121] This embodiment creatively applies Monte Carlo Dropout technology to the inference stage, enabling quantitative assessment of the uncertainty (standard deviation) of each model prediction. Uncertainty is incorporated as a key adjustment factor into the control strategy: when uncertainty is low, large and precise adjustments are made; when uncertainty is high, a conservative adjustment strategy is automatically adopted, limiting the adjustment magnitude. This "intelligent fault-tolerance" mechanism effectively avoids over-adjustment or system oscillations caused by inaccurate model predictions, greatly enhancing the operational safety and stability of the entire system and preventing potential damage to equipment.

[0122] This embodiment combines intelligent model prediction with a traditional PID controller. The neural network model is responsible for providing forward-looking, optimal target setpoints (frequency and power density), while the PID controller is responsible for tracking these target values ​​with high precision and disturbance resistance. This combination fully leverages the advantages of each, making the frequency and power density adjustment process of the ultrasonic transducer array both fast and smooth, effectively avoiding overshoot and oscillation, and ensuring the continuity and efficiency of the helium separation process.

[0123] Figure 5 This is a block diagram of a control system for a drilling helium gas acquisition device, as described in this exemplary embodiment. Figure 5 As shown, an exemplary embodiment of this disclosure provides a control system for a drilling helium acquisition device, comprising:

[0124] The acquisition unit is used to acquire monitoring data from the drilling helium acquisition device.

[0125] The construction unit is used to construct characteristic quantities reflecting fluid properties based on the monitoring data;

[0126] The training unit is used to train a prediction model for the control parameters of the acquisition device based on the monitoring data and the relationship between the feature quantities and the control parameters of the acquisition device.

[0127] The prediction unit is used to predict target control parameters based on the prediction model.

[0128] The control unit is used to dynamically control the helium collection device while drilling based on target control parameters.

[0129] Specifically, the monitoring data includes the viscosity, temperature, particle concentration, and pressure of the fluid to be helium extracted, and the construction unit includes:

[0130] The first construction module is used to construct a viscosity-temperature influence factor based on the viscosity and temperature of the fluid from which helium is to be extracted;

[0131] The second construction module is used to construct a particle pressure resistance factor based on the fluid particle concentration and pressure.

[0132] Specifically, the training unit includes:

[0133] A module is provided for building a neural network model, which includes an input layer, multiple hidden layers, and an output layer.

[0134] The adjustment module is used to take the monitoring data and feature quantities as input data of the neural network input layer and the control parameters as output data of the output layer. Based on the gradient descent method and the Adam optimizer, it continuously adjusts the weights of each connection in the neural network model to minimize the loss function of the neural network model and obtain the prediction model.

[0135] Specifically, the prediction unit includes:

[0136] The acquisition module is used to utilize Monte Carlo Dropout technology to perform multiple forward propagations and discard some network nodes during the inference phase of the prediction model, thereby obtaining the mean and standard deviation of the predicted values ​​of multiple control parameters.

[0137] The calculation module is used to dynamically calculate the target control parameters based on the mean and standard deviation of the predicted values, combined with a preset adjustment strategy and uncertainty tolerance.

[0138] Figure 6 This is a schematic diagram of the structure of a device according to an exemplary embodiment of this invention. Figure 6 As shown, corresponding to the above-described method for controlling a helium harvesting device while drilling, this disclosure also provides a device. Since the embodiment of this device is similar to the above-described method embodiment, the description is relatively simple; relevant details can be found in the description of the above-described method embodiment section. The device described below is merely illustrative. This device may include: a processor 1, a memory 2, a communication bus (i.e., the aforementioned device bus), and a lookup engine. The processor 1 and the memory 2 communicate with each other via the communication bus and communicate with external systems via a communication interface. The processor 1 can call logical instructions in the memory 2 to execute the method for controlling a helium harvesting device while drilling.

[0139] Furthermore, the logical instructions in the aforementioned memory 2 can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as memory chips, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0140] On the other hand, this disclosure also provides a processor-readable storage medium storing a computer program 3, which, when executed by a processor 1, is implemented to perform the drilling helium collection device control method provided in the above embodiments.

[0141] The processor-readable storage medium can be any available medium or data storage device that the processor 1 can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0142] The above are merely preferred embodiments of this disclosure. The scope of protection of this disclosure is not limited to the above embodiments. All technical solutions falling within the scope of this disclosure are protected. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this disclosure should be considered within the scope of protection of this disclosure.

Claims

1. A method for controlling a helium collection device while drilling, characterized in that, include: Acquire monitoring data from the drilling helium acquisition system; Based on the monitoring data, characteristic quantities reflecting fluid properties are constructed; Based on the relationship between the monitoring data and feature quantities and the control parameters of the acquisition device, a prediction model for the control parameters of the acquisition device is trained. Predict the target control parameters based on the prediction model; Dynamically regulate the helium collection device while drilling based on target control parameters; The monitoring data includes the viscosity, temperature, particle concentration, and pressure of the fluid to be extracted with helium. Based on the monitoring data, characteristic quantities reflecting the fluid properties are constructed, including: A viscosity-temperature influence factor is constructed based on the viscosity and temperature of the fluid to be helium extracted; Construct the particle pressure resistance factor based on fluid particle concentration and pressure; Based on the relationship between the monitoring data and feature quantities and the control parameters of the acquisition device, a prediction model for the control parameters of the acquisition device is trained, including: A neural network model is established, which includes an input layer, multiple hidden layers, and an output layer; The monitoring data and features are used as input data for the input layer of the neural network, and the control parameters are used as output data for the output layer. The weights of each connection in the neural network model are continuously adjusted based on gradient descent and the Adam optimizer to minimize the loss function of the neural network model and obtain the prediction model.

2. The method for controlling the helium collection device while drilling according to claim 1, characterized in that, Based on the prediction model, target control parameters are predicted, including: By using Monte Carlo Dropout technology, multiple forward propagations are performed during the inference phase of the prediction model, and some network nodes are discarded to obtain the mean and standard deviation of the predicted values ​​of multiple control parameters. Based on the mean and standard deviation of the predicted values, and combined with the preset adjustment strategy and uncertainty tolerance, the target control parameters are dynamically calculated.

3. A control system for a helium collection device while drilling, characterized in that, include: The acquisition unit is used to acquire monitoring data from the drilling helium acquisition device. The construction unit is used to construct characteristic quantities reflecting fluid properties based on the monitoring data; The training unit is used to train a prediction model for the control parameters of the acquisition device based on the monitoring data and the relationship between the feature quantities and the control parameters of the acquisition device. The prediction unit is used to predict target control parameters based on the prediction model. The control unit is used to dynamically control the helium acquisition device while drilling based on the target control parameters; The monitoring data includes the viscosity, temperature, particle concentration, and pressure of the fluid to be helium extracted. The building block includes: The first construction module is used to construct a viscosity-temperature influence factor based on the viscosity and temperature of the fluid from which helium is to be extracted; The second building module is used to build the particle pressure resistance factor based on the fluid particle concentration and pressure. Training units include: A module is provided for building a neural network model, which includes an input layer, multiple hidden layers, and an output layer. The adjustment module is used to take the monitoring data and feature quantities as input data of the neural network input layer and the control parameters as output data of the output layer. Based on the gradient descent method and the Adam optimizer, it continuously adjusts the weights of each connection in the neural network model to minimize the loss function of the neural network model and obtain the prediction model.

4. The control system for the drilling helium gas acquisition device according to claim 3, characterized in that, Prediction unit, including: The acquisition module is used to utilize Monte Carlo Dropout technology to perform multiple forward propagations and discard some network nodes during the inference phase of the prediction model, thereby obtaining the mean and standard deviation of the predicted values ​​of multiple control parameters. The calculation module is used to dynamically calculate the target control parameters based on the mean and standard deviation of the predicted values, combined with a preset adjustment strategy and uncertainty tolerance.

5. A device, characterized in that, include: Processor and memory; The memory is used to store computer programs, and the processor calls the computer programs stored in the memory to execute the drilling helium collection device control method according to any one of claims 1 to 2.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the processor to perform the drilling helium collection device control method according to any one of claims 1 to 2.

Images