A power supply system for oilfield high temperature and high pressure environment

Abstract

This invention discloses a power supply system for use in high-temperature and high-pressure environments in oilfields, comprising: an information acquisition module: acquiring initial temperature and pressure data under high-temperature and high-pressure conditions in oilfields, and obtaining downhole composite environment data through signal noise reduction preprocessing; a collaborative control module: based on the downhole composite environment data, constructing a thermal coupling optimization model by combining multi-physics field coupling with optimized neural network modeling, and outputting an acoustic-thermal energy collaborative control strategy; an optimization control module: extracting enhanced primary energy flow data based on the acoustic-thermal energy collaborative control strategy, and performing global optimization calculations on the enhanced primary energy flow data through particle swarm optimization algorithm combined with digital twin simulation to obtain the optimal power supply spectrum; and an intelligent power supply module: realizing intelligent power supply under high-temperature and high-pressure conditions in oilfields based on the optimal power supply spectrum.

Description

Technical Field

[0001] This invention relates to the field of intelligent power supply technology, and in particular to a power supply system for use in oilfields under high temperature and high pressure environments. Background Technology

[0002] The development of power supply technology in oilfields under high temperature and pressure environments has evolved from traditional power supply to intelligent power supply. Early systems primarily used a centralized power supply model, providing power to downhole equipment through surface substations. This model suffered from high energy loss and poor power supply stability. With technological advancements, distributed power supply systems gradually emerged, deploying multiple power supply nodes downhole, thus improving power reliability. Furthermore, with the rise of new technologies such as digital twins and artificial intelligence, power supply systems are moving towards intelligentization. Virtual models are being established to optimize power supply strategies, particularly in the area of ​​acoustic and thermal energy synergy. The abundant thermal and acoustic vibration energy downhole is being explored as auxiliary energy sources, with energy recovery achieved through thermoelectric conversion and piezoelectric effects, thereby improving the overall efficiency of the power supply system.

[0003] Existing power supply systems for high-temperature and high-pressure environments lack intelligent sensing capabilities for complex downhole environments in practical applications. They cannot dynamically adjust power supply parameters according to real-time operating conditions, and their synergistic utilization of acoustic and thermal energy is insufficient, resulting in low energy conversion efficiency. Furthermore, they lack effective multi-physics coupling control methods. In addition, the real-time interaction capability between digital twin models and physical systems is limited, making it difficult to achieve accurate prediction and optimization, which restricts the intelligent development of power supply systems. Therefore, this paper proposes a power supply system for high-temperature and high-pressure environments in oilfields. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention proposes the following technical solution:

[0005] A power supply system for use in high-temperature and high-pressure environments in oil fields, comprising:

[0006] Information acquisition module: Collects initial temperature and pressure data under high temperature and high pressure environment in oilfield, and obtains downhole composite environment data through signal noise reduction preprocessing operation;

[0007] Collaborative regulation module: Based on downhole composite environment data, a thermal coupling optimization model is constructed by combining multi-physics field coupling with optimized neural network modeling, and a collaborative regulation strategy for acoustic and thermal energy is output;

[0008] The thermal coupling optimization model consists of four parts: multiphysics coupling feature analysis, optimized neural network architecture design, neural network training, and model output strategy.

[0009] The optimized neural network architecture includes an input layer, a coupled feature layer, an optimization layer, and an output layer;

[0010] Optimized control module: Based on the acoustic and thermal energy coordinated regulation strategy, the enhanced native energy flow data is extracted, and the enhanced native energy flow data is globally optimized by combining the particle swarm optimization algorithm with digital twin simulation to obtain the optimal power supply spectrum;

[0011] Intelligent power supply module: Based on the optimal power supply map, it realizes intelligent power supply in the high temperature and high pressure environment of oil fields.

[0012] The process of obtaining downhole composite environmental data is as follows:

[0013] The initial temperature and pressure data were collected to form a raw dataset. This raw dataset... An adaptive wavelet threshold denoising method is used for processing, and the selected... Wavelet basis functions were used, and the decomposition level was set to 3 levels. After decomposition, wavelet coefficients at each scale were obtained. ,in Indicates the decomposition scale;

[0014] Then, the adaptive thresholds at each scale are calculated. Based on adaptive threshold Thresholding processing is performed on wavelet coefficients at various scales to obtain downhole composite environment data. .

[0015] The implementation process of multiphysics coupling characteristic analysis is as follows:

[0016] Downhole composite environment data Physical environment feature extraction is performed, including sound field features. Thermal characteristics ;

[0017] The sound field characteristics By using downhole composite environment data Acoustic signals are acquired and their time-domain amplitude characteristics are calculated. Then, the time-domain acoustic signals are converted into frequency-domain representations using Fourier transform.

[0018] The thermal field characteristics By analyzing downhole composite environment data Temperature data from the sensor array Spatial interpolation is used to directly obtain the data, where... Spatial coordinates:

[0019] sound field characteristics Thermal characteristics The input vector of the quantized neural network is represented as an environmental feature vector. .

[0020] The implementation process for optimizing neural network architecture design is as follows:

[0021] The input layer receives the environmental feature vector after multiphysics coupling feature analysis. The number of nodes equals the extracted feature dimension;

[0022] The coupling feature layer processes the spatial coupling sub-features of the thermal field through a convolutional neural network. Temporal coupling sub-features of the sound field are processed through a recurrent neural sub-network. Then, the coupling delay time between the sound field and the thermal field is obtained and concatenated with the sub-features to output the sound-thermal coupling feature. ;

[0023] The optimization layer receives acoustic-thermal coupling features. And by introducing physical constraint loss Physical constraint loss Including energy conservation constraint losses and the constraint loss of the second law of thermodynamics Based on physical constraint loss The network parameters are updated using a combination of gradient descent and physical regularization.

[0024] The output layer converts the acoustic-thermal coupling characteristics through linear transformation. Feature dimensions mapped to Obtain the initial control strategy vector And will initially regulate the strategy vector Each parameter is mapped to the adjustable range of the actual device to obtain a coordinated control strategy for acoustic and thermal energy. .

[0025] The process of obtaining the coupling delay time between the sound field and the thermal field is as follows:

[0026] Spatial Coupler Features Based on Thermal Field Temporal Coupler Features of Sound Field Through cross-correlation function The location of the maximum value yields the coupling delay time of the acoustic-thermal field. .

[0027] The training process for optimizing a neural network is as follows:

[0028] Constructing training data for optimized neural networks, including strategies for coordinated regulation of acoustic and thermal energy. Combined with historical downhole environmental data Then, the data is normalized, and the total loss function is obtained by combining the mean squared error and the physical constraint loss. ;

[0029] Then adopt The optimizer trains the neural network, determines the weights and biases of the neural network, and completes the training.

[0030] Finally, after optimizing the neural network architecture, a thermally coupled optimization model is obtained. .

[0031] The process of enhancing the acquisition of native energy flow data is as follows:

[0032] Analysis of the synergistic regulation strategy of acoustic and thermal energy The core control parameters included are acoustic generator power and heat exchanger flow rate. Downhole composite environmental data is retrieved, and the original acoustic energy flow component is enhanced and corrected based on the acoustic generator power. The original thermal energy flow component is directionally enhanced based on the heat exchanger flow rate. The enhanced acoustic energy flow component and the thermal energy flow component are then stitched together according to well depth and time dimensions to obtain the enhanced original energy flow data. .

[0033] The process of performing simulation-based global optimization calculations on enhanced native energy flow data is as follows:

[0034] Based on the particle swarm optimization algorithm, let the objective function of the particle swarm optimization algorithm be: ,in, Based on the basic power supply parameter vector, define the constraints for the optimization process. These constraints include requirements for the simulation output. Approaching the target And find the Minimum power supply parameters;

[0035] Based on the particle swarm optimization algorithm, the position of each particle is mapped to a set of power supply parameter vectors. , The particle index is defined as the position range limited to the rated parameter range of the oilfield power supply equipment, and the particle velocity is initialized to... The parameter range width is increased by times, and then digital twin simulation optimization combined with particle swarm optimization algorithm is performed;

[0036] The power supply parameter vector corresponding to the globally optimal position is obtained based on digital twin simulation optimization. The globally optimal power supply parameters are set according to The system categorizes power supply based on two dimensions: well depth and time, resulting in an optimal power supply map that includes the best power supply parameters for each well segment throughout the entire time period. .

[0037] The process of digital twin simulation optimization combined with particle swarm optimization algorithm is as follows:

[0038] Based on the actual well section layout and power supply line parameters of the oilfield, a system is constructed that integrates with the physical environment. The digital twin model of the mapping, with the input being the first... Power supply parameter vector of each particle The output is the first... Simulated energy flow of individual particles ;

[0039] Then iteratively update the particle positions, setting each updated particle position as... Inputting the digital twin model yields the simulated energy flow corresponding to each updated particle position. And calculate its relationship with Authenticity error If the accuracy is incorrect If the error is less than the preset error, then calculate the target function value corresponding to that particle. Otherwise, update the particle position;

[0040] The iteration stops when the number of iterations reaches a preset maximum value. At this point, the power supply parameter vector corresponding to the globally optimal position is the optimal power supply parameter. .

[0041] The present invention has the following beneficial effects:

[0042] 1. By optimizing neural network modeling and introducing energy conservation constraints and the second law of thermodynamics constraints, and by calibrating the model output with physical rules through a loss function, the ineffective strategies of traditional pure data fitting models, such as energy non-conservation and reverse heat transfer, are avoided. At the same time, by following the energy distribution logic in the directional enhancement process of acoustic and thermal energy, both efficient energy utilization and ensuring that the equipment always operates within a safe physical threshold are achieved. Compared with traditional solutions, the probability of equipment failure due to energy imbalance is reduced.

[0043] 2. By using a synergistic control strategy for acoustic and thermal energy and an optimal power supply spectrum, the synergistic utilization of acoustic and thermal energy is achieved. The energy loss caused by downhole fluid disturbance is reduced by enhancing the acoustic energy flow, and the heat dissipation efficiency of equipment is optimized by the directional distribution of thermal energy flow, thereby improving the utilization rate of power supply energy in the high temperature and high pressure environment of the oilfield. Furthermore, the well section-time dimension parameter matrix of the optimal power supply spectrum enables precise scheduling of power supply resources. Attached Figure Description

[0044] Figure 1 This is a system block diagram of a power supply system for use in oilfields under high temperature and high pressure environments, as proposed in this invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Example 1

[0047] like Figure 1 As shown, the present invention proposes a power supply system for use in high-temperature and high-pressure environments in oil fields, comprising:

[0048] Information acquisition module: Collects initial temperature and pressure data under high temperature and high pressure environment in oilfield, and obtains downhole composite environment data through signal noise reduction preprocessing operation;

[0049] Deploy temperature and pressure sensors; the temperature sensor is used to collect time data. Temperature data is collected internally, and pressure sensors are used to acquire time data. Internal pressure data is collected, and these collected data are integrated to obtain the raw dataset. ;

[0050] Specifically, the original dataset It forms the basis for subsequent processing, providing the original input data for subsequent signal noise reduction preprocessing;

[0051] For the original dataset Noise reduction is performed by using filtering algorithms to remove noise from the data, targeting the original dataset. An adaptive wavelet threshold denoising method is used for processing. First, wavelet decomposition is performed, and then... Wavelet basis functions, decomposition level set to... Layers, after decomposition, yield wavelet coefficients at each scale. ,in Indicates the decomposition scale;

[0052] Then, the adaptive thresholds at each scale are calculated. ,in, For the first The noise standard deviation of wavelet coefficients at different scales. This represents the number of sampling points for the sound wave signal.

[0053] Based on adaptive threshold Thresholding is applied to wavelet coefficients at various scales. Wavelet coefficients with amplitudes greater than the threshold are shrunk, while those with amplitudes less than or equal to the threshold are set to zero. This approach preserves the main characteristics of the signal while effectively eliminating noise. After thresholding, wavelet reconstruction is performed on the processed coefficients using the same wavelet basis functions as during decomposition. Inverse wavelet transform is used to reconstruct the denoised acoustic signal from the processed coefficients. The reconstructed signal retains the main characteristics of the original signal while significantly reducing noise levels, thus obtaining downhole composite environment data. .

[0054] Collaborative regulation module: Based on downhole composite environment data, a thermal coupling optimization model is constructed by combining multi-physics field coupling with optimized neural network modeling, and a collaborative regulation strategy for acoustic and thermal energy is output;

[0055] Based on downhole composite environment data By combining multi-physics coupling optimization neural network modeling methods, a thermal coupling optimization model is constructed. In this process, the first step is to analyze the downhole composite environment data. The characteristics and interrelationships of the acoustic and thermal energy fields contained within are analyzed, and then a model is established using a multi-physics coupled optimized neural network. The input is assumed to be downhole composite environment data. The output is a coordinated control strategy of acoustic and thermal energy. ;

[0056] The thermal coupling optimization model consists of four parts: multiphysics coupling feature analysis, optimized neural network architecture design, neural network training, and model output strategy.

[0057] The implementation process of multiphysics coupling characteristic analysis is as follows:

[0058] For downhole complex environment data Physical environment features are extracted based on the physical field coupling mechanism, and the physical field coupling mechanism is used to analyze downhole composite environment data. In the downhole environment, dynamic coupling relationships exist, such as the coupling delay time and energy conversion efficiency of the acoustic and thermal fields. Therefore, the physical environment features extracted by multiphysics coupling feature analysis include acoustic field features. and thermal field characteristics ;

[0059] Specifically, the sound pressure amplitude is obtained from downhole composite environmental data. Acoustic signals are acquired directly by collecting them and calculating their time-domain amplitude characteristics, and then combined with Fourier transform. The time-domain acoustic signal is converted into a frequency-domain representation for direct acquisition.

[0060] Thermal characteristics That is, the spatial distribution of the temperature field, through the analysis of downhole composite environment data. Temperature data from downhole thermocouples, infrared thermometers, and other sensor arrays. ( Directly obtain the coordinates (using spatial coordinates) through spatial interpolation:

[0061] These features are quantized into the input vector of the neural network, represented as the environmental feature vector. , To obtain data from complex environments The sound field features extracted from it. To obtain data from complex environments The thermal field features extracted from them;

[0062] The implementation process for optimizing neural network architecture design is as follows:

[0063] Optimizing neural network architecture includes a neural network architecture consisting of an input layer, a coupled feature layer, an optimization layer, and an output layer;

[0064] Input layer implementation process:

[0065] Receive environmental feature vectors after multiphysics coupling feature analysis The number of nodes equals the extracted feature dimension;

[0066] Implementation process of coupled feature layer:

[0067] The coupling feature layer processes the spatial coupling features of the thermal field through a convolutional neural sub-network, processes the temporal coupling features of the acoustic field through a recurrent neural sub-network, and then strengthens the weights of key acoustic-thermal coupling features through an attention mechanism.

[0068] Specifically, the process by which convolutional neural subnetworks process the spatial coupler features of the thermal field is as follows:

[0069] Convolutional neural subnetworks extract environmental feature vectors The sub-features corresponding to the thermal features are separated and reshaped into a spatial distribution matrix; for example, if the thermal feature is the temperature gradient at different depths of the well section, it is reshaped into [number of well section depths, ... A matrix of ];

[0070] Then perform the convolution operation, setting the kernel size to... The spatial dimension and step size for adapting to well depth are ,pass The convolution operation extracts spatial correlation features from the temperature field. The convolution output is then subjected to a nonlinear transformation using the ReLU activation function, followed by max pooling to compress the feature dimension, resulting in spatially coupled sub-features of the thermal features. ;

[0071] The process by which a recurrent neural subnetwork processes the temporal coupler features of a sound field is as follows:

[0072] Recurrent neural subnetworks from environmental feature vectors The sub-vectors corresponding to the acoustic features are extracted and reshaped into a temporal sequence.

[0073] For example, if the acoustic characteristics are the sound wave energy densities at different times, then the reshaping is [time steps, ] sequence;

[0074] Then perform a loop to set the number of hidden layer nodes. ,pass The temporal memory ability of sound waves allows for the extraction of temporal propagation characteristics, such as the attenuation pattern of sound wave energy at different times. The hidden layer output at the last time step serves as a temporally coupled sub-feature of the acoustic features. ;

[0075] Then, the coupling delay time between the acoustic and thermal fields is obtained. Spatial coupling sub-features with thermal characteristics Temporal coupling sub-features with acoustic features Perform feature concatenation, where:

[0076] The coupling delay time is determined by analyzing the time series correlation between the acoustic field and the thermal field, based on the spatial coupler characteristics of the thermal field. Temporal Coupler Features of Sound Field Through cross-correlation function The location of the maximum value yields the coupling delay time of the acoustic-thermal field. Cross-correlation function Represented as:

[0077] ;

[0078] in, The mean value of the sound field signal. Let E[⋅] be the mean of the thermal field signal, and E[⋅] be the expectation operator. Let V be the variance of the sound field signal. Let Variance be the variance of the thermal field signal, when When the maximum value is reached, the corresponding τ is the coupling delay time;

[0079] Spatial coupling of thermal features with sub-features Temporal coupling sub-features of acoustic features With coupling delay time Feature concatenation is performed to obtain the acoustic-thermal coupling features of the output of the coupling layer. ;

[0080] The implementation process of the optimization layer is as follows:

[0081] Optimization layer receives acoustic-thermal coupling characteristics Furthermore, by introducing physical constraint losses, the output of the optimized neural network is made to satisfy physical laws. These physical constraint losses include energy conservation constraint losses and second law of thermodynamics constraint losses. The energy conservation constraint loss function... Represented as the rate of all input heats With all output heat rates The goal of this loss is to minimize the difference between input and output energy, ensuring that the system follows the energy balance of a thermodynamic system.

[0082] The second law of thermodynamics constrains the loss function The entropy value of the input energy and the entropy of the output energy after regulation To maximize the loss;

[0083] The final physical constraint loss Represented as the energy conservation constraint loss function The second law of thermodynamics constrains the loss function sum;

[0084] Then, the optimization layer is based on physical constraint loss. The network parameters are updated using a gradient descent combined with physical regularization. Calculate the gradient of each layer's parameters, and adjust the parameters according to the gradient descent rule. or If the threshold is exceeded, the corresponding loss weight is increased to strengthen the effect of physical constraints. The change amount is less than 10 consecutive iterations ,and , Training stops when all values ​​are less than the threshold.

[0085] The implementation process of the output layer is as follows:

[0086] The output layer uses linear transformation to convert the acoustic-thermal coupling characteristics. Feature dimensions mapped to That is, the number of nodes in the output layer is set as the dimension of the control parameter. To meet the dimensions of the control parameters Equal to the number of sound control parameters Such as the adjustment dimensions of the sound wave generator power and the number of thermal control parameters. Such as the sum of the dimensions of flow regulation in heat exchangers;

[0087] Specifically, the output layer inputs the output features into the fully connected layer, and maps the feature dimensions to a linear transformation. The initial regulation strategy vector is obtained. Then, the initial control strategy vector Each parameter is mapped to the adjustable range of the actual equipment, and the control strategy vector after range constraint is the acoustic-thermal energy coordinated control strategy. This includes parameters such as the power of the sound wave generator, the flow rate of the heat exchanger, and the energy distribution ratio.

[0088] For example: power mapping of the sound wave generator to... To the rated power, the heat exchanger flow rate is mapped to the minimum flow rate and the maximum flow rate;

[0089] The process of optimizing neural network training is as follows:

[0090] The training data for optimizing the neural network includes acoustic-thermal energy coordinated control strategies and historical downhole composite environment data. Historical monitoring data of downhole complex environment data were collected, as well as downhole complex environment data under different operating conditions. A series of acoustic and thermal energy coordinated regulation strategies for output A training set is constructed, and then the data is normalized and combined with mean squared error. And physical constraint loss, define the total loss for optimizing a neural network. for The sum of physical constraint losses, To optimize the control strategy of neural network output, For real strategy, This represents the mean square error between the output control strategy and the actual strategy.

[0091] use The optimizer sets the learning rate, batch size, and number of training epochs, and divides the training set proportionally. and verification set During training, the loss on the validation set is monitored in real time. Training is stopped when the validation set loss no longer decreases. After training, the generalization ability of the model is verified using historical data not used in training. The prediction error on the test set is calculated, and the physical rationality of the adjustment strategy is evaluated. If the model performance meets the requirements, the test set is considered acceptable. If the weights and biases of the neural network are less than the preset threshold and the physical constraint loss is less than the preset threshold, then the training is completed.

[0092] Finally, after optimizing the neural network architecture, a thermally coupled optimization model is constructed. The formula is expressed as: ;

[0093] Specifically, this process achieves a thermally coupled optimization model through a closed-loop workflow of physical field analysis, neural network architecture design, training optimization, and model building. The construction of the multi-physics coupled optimization neural network, by capturing the nonlinear coupling relationship between the acoustic and thermal fields and introducing physical constraint losses, ensures that the output regulation strategy not only conforms to physical laws but also effectively optimizes energy utilization. This model provides a key collaborative regulation strategy for the subsequent optimization control module, supporting the intelligent operation of the entire collaborative regulation system.

[0094] Optimized control module: Based on the acoustic and thermal energy coordinated regulation strategy, the enhanced native energy flow data is extracted, and the enhanced native energy flow data is globally optimized by combining the particle swarm optimization algorithm with digital twin simulation to obtain the optimal power supply spectrum;

[0095] The process of enhancing native energy flow data acquisition is as follows:

[0096] Synergistic Regulation Strategy of Acoustic and Thermal Energy Output Based on Synergistic Regulation Module By extracting operators Extracting enhanced native energy flow data ;

[0097] Specifically, extraction operator The acquisition process is as follows:

[0098] First, analysis The core control parameters included are the power of the sound wave generator. Heat exchanger flow rate ;

[0099] Secondly, the original downhole composite environment data is retrieved, and the power of the acoustic generator is used as a reference. For the original sound wave energy flow components Enhancement is carried out based on the heat exchanger flow rate. For the original thermal energy flow component Enhance;

[0100] Specifically, depending on the power of the sound wave generator The process of enhancing the original sound wave energy flow components is as follows:

[0101] Changes in the power of the acoustic generator directly affect the energy intensity of the downhole acoustic waves. Multiplying this by the original sound wave energy flow component yields the enhanced sound wave energy flow component. ;

[0102] Based on heat exchanger flow rate The process of directionally enhancing the original thermal energy flow component is as follows:

[0103] From the original downhole composite environmental data, obtain the original thermal energy flow components of different regions within the corresponding well section, and when the heat exchanger flow rate... When a well section is inclined towards a certain area, the heat exchanger increases the flow rate of the heat medium in that area, thereby improving the heat transfer efficiency of that area. Taking multiple sub-regions within a well section as an example, if the flow rate allocated to a certain sub-region by the heat exchanger is... Then the enhanced thermal energy flow component of this sub-region The original heat energy flow component and the proportion of heat exchanger flow allocated to sub-region i. The product of these two factors is used to achieve a directional and concentrated enhancement of thermal energy flow to the target area, while ensuring the conservation of the total thermal energy flow throughout the entire well section.

[0104] Finally, the enhanced acoustic energy flow component and the thermal energy flow component are stitched together according to the well depth and time dimension to obtain the enhanced native energy flow data. ;

[0105] Specifically, It includes synergistic rules for acoustic energy-driven energy transfer and thermal energy-regulated energy distribution, such as acoustic vibration enhancing fluid mixing and thermal gradient guiding directional energy transfer, and extraction operators. These rules will be parsed to perform feature enhancement on the original energy flow data, resulting in enhanced original energy flow data. It has clearer energy flow characteristics, providing high-quality basic input for subsequent global optimization and avoiding the optimization algorithm from getting stuck in local optima or invalid search due to the ambiguity of the original data;

[0106] Particle Swarm Optimization Algorithm Combining digital twin simulation to enhance native energy flow data Perform global optimization calculations to obtain the optimal power supply pattern. ;

[0107] Let the objective function of the particle swarm optimization algorithm be... ,in, The basic power supply parameter vector includes the power supply voltage, current, and power supply for each well section. The output after digital twin simulation is the current simulation output. Then the expression for the optimization process is:

[0108]

[0109] in, This section represents the constraints and requirements for the simulation output. Get as close to the target as possible To ensure the optimization solution is feasible in real-world scenarios, Indicates finding Minimum power supply parameters The process;

[0110] Then define the power cooperative objective function. The power coordination objective function needs to simultaneously reflect power supply efficiency and stability, taking into account the power supply requirements of the high-temperature and high-pressure environment of the oilfield, including the total system loss rate target and the maximum voltage deviation target obtained by direct measurement. Wherein, the total system loss rate target is expressed as the first... Power loss in the well section Table and total power supply The ratio;

[0111] Based on the particle swarm optimization algorithm, the position of each particle is mapped to a set of power supply parameter vectors. , The particle index is defined as the position range limited to the rated parameter range of the oilfield power supply equipment, and the particle velocity is initialized to... The parameter interval width is increased by a factor of 1, and then the digital twin simulation optimization algorithm is used in conjunction with the particle swarm optimization algorithm.

[0112] Based on the actual well section layout and power supply line parameters of the oilfield, a system is constructed that integrates with the physical environment. A mapped digital twin model, the input of which is the first... Power supply parameter vector of each particle The output is a simulated energy flow. It includes the acoustic energy flow and thermal energy flow components of each well section, and the dimensions are... Consistent;

[0113] Then iteratively update the particle positions, setting each updated particle position as... Input the digital twin model to obtain the simulated energy flow And calculate its relationship with The accuracy error is expressed by the formula:

[0114] ;

[0115] in, Due to the error in authenticity, For particle indexing, For energy flow data dimensions;

[0116] If the truth is inaccurate If the error is less than the preset error, then calculate the target function value corresponding to that particle. Otherwise, update the particle position;

[0117] When the number of iterations reaches the preset maximum value, or the change in the global optimum is continuous The second iteration is less than When the iteration stops, the power supply parameter vector corresponding to the global optimal position at this time is obtained. This refers to the optimal power supply parameters for each well section at different times, which are organized into a matrix according to the well section depth and time dimension:

[0118] The globally optimal power supply parameters are set according to The well section is categorized based on two dimensions: depth and time, clearly defining the depth of each well section. At different times The corresponding power supply parameters are then used for matrix mapping, constructing a two-dimensional matrix with well depth as the row and time as the column: the row index corresponds to the well depth. Column index corresponding time The matrix elements are the well section. At any moment Optimal power supply parameters The matrix is ​​visualized according to the correlation between well section, time, and parameters, forming an optimal power supply map that includes the optimal power supply parameters for each well section throughout the entire time period. .

[0119] Intelligent power supply module: realizes intelligent power supply in oilfield high temperature and high pressure environment based on optimal power supply map;

[0120] Optimal power supply diagram The corresponding elements are parsed as time-parameter commands for each power supply unit, according to well depth. Match the corresponding downhole power supply unit and extract the data at each time point from the map. The voltage and power parameters corresponding to the power supply unit are used to generate a real-time control command set, which is then sent to the controller of the power supply unit.

[0121] The power supply unit performs real-time parameter control based on the issued instructions and the real-time status of the high temperature and high pressure environment of the oilfield. Through the frequency converter, the output voltage and power of the power supply unit are adjusted to the voltage and power corresponding to the instructions.

[0122] Ultimately, a closed loop is achieved encompassing power supply units, data acquisition, and graph updates, enabling real-time collection of actual output parameters (voltage, current, power), equipment operating temperature, and enhanced native energy flow data from each power supply unit. Compare the deviation value with the target parameters of the optimal power supply pattern, and provide deviation feedback: if the deviation exceeds the threshold, the voltage deviation is greater than... The deviation data is sent back to the optimization control module, triggering a local map update, and the short-term optimal parameters for the well section are recalculated based on the current environmental data.

[0123] When the oilfield environment undergoes continuous changes, the linkage information acquisition module updates the original data, re-executes the collaborative regulation-optimization control process, generates an updated optimal power supply map, and realizes dynamic iteration of the power supply strategy.

[0124] In the application, several formulas are calculated by removing dimensions and taking their numerical values. The formulas are established by collecting a large amount of data and simulating the most recent real situation. Some coefficients or weights in the formulas are set by those skilled in the art according to the actual situation, so they will not be elaborated here.

[0125] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0126] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A power supply system for use in high-temperature and high-pressure environments in oil fields, characterized in that, include: Information acquisition module: Collects initial temperature and pressure data under high temperature and high pressure environment in oilfield, and obtains downhole composite environment data through signal noise reduction preprocessing operation; Collaborative regulation module: Based on downhole composite environment data, a thermal coupling optimization model is constructed by combining multi-physics field coupling with optimized neural network modeling, and a collaborative regulation strategy for acoustic and thermal energy is output; The thermal coupling optimization model consists of four parts: multiphysics coupling feature analysis, optimized neural network architecture design, neural network training, and model output strategy. The optimized neural network architecture includes an input layer, a coupled feature layer, an optimization layer, and an output layer; Optimized control module: Based on the acoustic and thermal energy coordinated regulation strategy, the enhanced native energy flow data is extracted, and the enhanced native energy flow data is globally optimized by combining the particle swarm optimization algorithm with digital twin simulation to obtain the optimal power supply spectrum; Intelligent power supply module: realizes intelligent power supply in oilfield high temperature and high pressure environment based on optimal power supply map; The implementation process of multiphysics coupling characteristic analysis is as follows: Downhole composite environment data Physical environment feature extraction is performed, including sound field features. Thermal characteristics ; The sound field characteristics By using downhole composite environment data Acoustic signals are acquired and their time-domain amplitude characteristics are calculated. Then, the time-domain acoustic signals are converted into frequency-domain representations using Fourier transform. The thermal field characteristics By analyzing downhole composite environment data Temperature data from the sensor array Spatial interpolation is used to directly obtain the data, where... Spatial coordinates: sound field characteristics Thermal characteristics The input vector of the quantized neural network is represented as an environmental feature vector. ; The implementation process for optimizing neural network architecture design is as follows: The input layer receives the environmental feature vector after multiphysics coupling feature analysis. The number of nodes equals the extracted feature dimension; The coupling feature layer processes the spatial coupling sub-features of the thermal field through a convolutional neural network. Temporal coupling sub-features of the sound field are processed through a recurrent neural sub-network. Then, the coupling delay time between the sound field and the thermal field is obtained and concatenated with the sub-features to output the sound-thermal coupling feature. ; The optimization layer receives acoustic-thermal coupling features. And by introducing physical constraint loss Physical constraint loss Including energy conservation constraint losses and the constraint loss of the second law of thermodynamics Based on physical constraint loss The network parameters are updated using a combination of gradient descent and physical regularization. The output layer uses linear transformation to convert the acoustic-thermal coupling characteristics. Feature dimensions mapped to Obtain the initial control strategy vector And will initially regulate the strategy vector Each parameter is mapped to the adjustable range of the actual device to obtain a coordinated control strategy for acoustic and thermal energy. .

2. The power supply system for use in high-temperature and high-pressure environments in oil fields according to claim 1, characterized in that, The process of obtaining downhole composite environmental data is as follows: The initial temperature and pressure data were collected to form a raw dataset. This raw dataset... An adaptive wavelet threshold denoising method is used for processing, and the selected... Wavelet basis functions were used, and the decomposition level was set to 3 levels. After decomposition, wavelet coefficients at each scale were obtained. ,in Indicates the decomposition scale; Then, the adaptive thresholds at each scale are calculated. Based on adaptive threshold Thresholding processing is performed on wavelet coefficients at various scales to obtain downhole composite environment data. .

3. A power supply system for use in high-temperature and high-pressure environments in oil fields according to claim 1, characterized in that, The process of obtaining the coupling delay time between the sound field and the thermal field is as follows: Spatial Coupler Features Based on Thermal Field Temporal Coupler Features of Sound Field Through cross-correlation function The location of the maximum value yields the coupling delay time of the acoustic-thermal field. .

4. A power supply system for use in high-temperature and high-pressure environments in oil fields according to claim 3, characterized in that, The optimization process for training a neural network is as follows: Constructing training data for optimized neural networks, including strategies for coordinated regulation of acoustic and thermal energy. Combined with historical downhole environmental data Then, the data is normalized, and the total loss function is obtained by combining the mean squared error and the physical constraint loss. ; Then adopt The optimizer trains the neural network, determines the weights and biases of the neural network, and completes the training. Finally, after optimizing the neural network architecture, a thermally coupled optimization model is obtained. .

5. A power supply system for use in high-temperature and high-pressure environments in oil fields according to claim 4, characterized in that, The process of enhancing the acquisition of native energy flow data is as follows: Analysis of the coordinated regulation strategy of acoustic and thermal energy The core control parameters included are acoustic generator power and heat exchanger flow rate. Downhole composite environmental data is retrieved, and the original acoustic energy flow component is enhanced and corrected based on the acoustic generator power. The original thermal energy flow component is directionally enhanced based on the heat exchanger flow rate. The enhanced acoustic energy flow component and the thermal energy flow component are then stitched together according to well depth and time dimensions to obtain the enhanced original energy flow data. .

6. A power supply system for use in high-temperature and high-pressure environments in oil fields according to claim 5, characterized in that, The process of performing simulation-based global optimization calculations on enhanced native energy flow data is as follows: Based on the particle swarm optimization algorithm, let the objective function of the particle swarm optimization algorithm be: ,in, Based on the basic power supply parameter vector, define the constraints for the optimization process. These constraints include requirements for the simulation output. Approaching the target And find the Minimum power supply parameters; Based on the particle swarm optimization algorithm, the position of each particle is mapped to a set of power supply parameter vectors. , The particle index is defined as the position range limited to the rated parameter range of the oilfield power supply equipment, and the particle velocity is initialized to... The parameter range width is increased by times, and then digital twin simulation optimization combined with particle swarm optimization algorithm is performed; The power supply parameter vector corresponding to the globally optimal position is obtained based on digital twin simulation optimization. The globally optimal power supply parameters are set according to The system categorizes power supply based on two dimensions: well depth and time, resulting in an optimal power supply map that includes the best power supply parameters for each well segment throughout the entire time period. .

7. A power supply system for use in high-temperature and high-pressure environments in oil fields according to claim 6, characterized in that, The process of digital twin simulation optimization combined with particle swarm optimization algorithm is as follows: Based on the actual well section layout and power supply line parameters of the oilfield, a system is constructed that integrates with the physical environment. The digital twin model of the mapping, with the input being the first... Power supply parameter vector of each particle The output is the first... Simulated energy flow of individual particles ; Then iteratively update the particle positions, setting each updated particle position as... Inputting the digital twin model yields the simulated energy flow corresponding to each updated particle position. And calculate its relationship with Authenticity error If the accuracy is incorrect If the error is less than the preset error, then calculate the target function value corresponding to that particle. Otherwise, update the particle position; The iteration stops when the number of iterations reaches a preset maximum value. At this point, the power supply parameter vector corresponding to the globally optimal position is the optimal power supply parameter. .

Images