A Dynamic Analysis Method and System for Well Site Fracturing Based on the Internet of Things

Abstract

This invention relates to the field of well site fracturing technology and discloses a dynamic analysis method for well site fracturing based on the Internet of Things, comprising: Step S1: Constructing a multi-level sensor network at the well site for real-time acquisition of fracturing equipment operating parameters, downhole reservoir dynamic response data, fracturing fluid performance parameters, and well site surrounding environment monitoring data to form multi-source data; Step S2: Preprocessing the acquired multi-source data through edge computing nodes to obtain preprocessed data, and uploading the preprocessed data to a cloud analysis platform via an encrypted transmission protocol, wherein the cloud analysis platform has a well site fracturing digital twin model; Step S3: The cloud analysis platform performs correlation calculations on the preprocessed data and the digital twin model through a multi-dimensional analysis model to generate a comprehensive evaluation result of fracture propagation dynamic characteristics, reservoir conductivity, and fracturing equipment operating status; Step S4: Generating control commands based on the comprehensive evaluation result to dynamically adjust fracturing parameters.

Description

Technical Field

[0001] This invention relates to the field of well site fracturing technology, specifically to a dynamic analysis method and system for well site fracturing based on the Internet of Things. Background Technology

[0002] Well site fracturing is a key production enhancement technology for low-permeability reservoirs in oil and gas field development. It involves forcibly injecting fracturing fluid into the underground reservoir using high-pressure equipment to create artificial fractures and keep them open, thereby improving the flow channels for oil and gas and increasing the recovery rate.

[0003] Traditional well site fracturing dynamic analysis mostly relies on local data monitoring and manual experience control. Due to incomplete data coverage and lack of real-time feedback in control, the fracturing effect deviates significantly from the design expectations.

[0004] In view of this, the present invention is hereby proposed. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a dynamic analysis method and system for well site fracturing based on the Internet of Things (IoT). This method solves the problem that traditional real-time pressure prediction methods for fracturing wellheads are insufficient to characterize the pressure evolution under complex operating conditions, leading to the continuous accumulation of prediction errors as the construction progresses, thus affecting the accuracy of the predictions.

[0006] Specifically, the following technical solution was adopted:

[0007] A dynamic analysis method for well site fracturing based on the Internet of Things includes:

[0008] Step S1: Construct a multi-level sensor network at the well site to collect real-time operating parameters of fracturing equipment, dynamic response data of downhole reservoirs, performance parameters of fracturing fluid, and monitoring data of the surrounding environment of the well site, forming multi-source data;

[0009] Step S2: The collected multi-source data is preprocessed through edge computing nodes to obtain preprocessed data, and the preprocessed data is uploaded to the cloud analysis platform via an encrypted transmission protocol. The cloud analysis platform has a well site fracturing digital twin model.

[0010] Step S3: The cloud-based analysis platform performs correlation calculations on the preprocessed data and digital twin model through a multi-dimensional analysis model to generate a comprehensive evaluation result of the dynamic characteristics of fracture propagation, reservoir conductivity, and fracturing equipment operation status. The multi-dimensional analysis model includes a fracture propagation model, a proppant migration model, and a reservoir productivity prediction model. The fracture propagation model is constructed based on fracture mechanics theory and is used to calculate the dynamic characteristics of fracture propagation. The proppant migration model calculates the migration law of proppant in the fracture based on the proppant concentration and fracturing fluid flow rate in the multi-source data. The reservoir productivity prediction model predicts the productivity after fracturing based on the reservoir parameters in the multi-source data.

[0011] Step S4: Generate control instructions based on the comprehensive evaluation results, and dynamically adjust the fracturing parameters.

[0012] As an optional embodiment of the present invention, in a well site fracturing dynamic analysis method based on the Internet of Things, step S2 involves preprocessing the collected multi-source data using edge computing nodes to obtain preprocessed data, including:

[0013] Step S201: Using the formula Identifying outliers in multi-source data, among which For a single sensor data point, This represents the average value of data within the same time period. The standard deviation is used to identify data that meet this condition as outliers, and the Laida criterion is used to remove outliers.

[0014] Step S202: Align the time axis of the multi-source data after removing outliers, and convert the acquisition timestamps of different sensors into relative times with the start time of the fracturing pump as the reference zero point.

[0015] Step S203: Normalization is used to eliminate the influence of dimensions, and preprocessed data is obtained.

[0016] As an optional embodiment of the present invention, in a well site fracturing dynamic analysis method based on the Internet of Things, the construction of a well site fracturing digital twin model by the cloud analysis platform in step S2 includes: constructing a physical field simulation model containing stress field, seepage field and temperature field using finite element analysis software based on well site three-dimensional geological modeling data, well completion parameters and historical fracturing data, so as to form a well site fracturing digital twin model. The update frequency of the well site fracturing digital twin model is consistent with the sensor data acquisition frequency of the multi-level sensor network.

[0017] As an optional embodiment of the present invention, in a well site fracturing dynamic analysis method based on the Internet of Things, the fracture propagation model is determined by the formula... Calculate the type I fracture toughness, where Type I fracture toughness, Geometric factor For reservoir geostress, For the crack half-length, the input data includes strain values ​​collected by distributed fiber optic strain sensors in the multi-level sensor network, source location data collected by microseismic detectors, and geostress logging data.

[0018] The proppant transport model is derived from the formula. Describe the proppant transport behavior, among which For proppant concentration, For time, This refers to the fracturing fluid flow rate. The coordinates are along the crack length direction. The diffusion coefficient is... The input data, representing the proppant settling coefficient, includes fracturing fluid flow sensor data, proppant concentration detector data, and the fracture half-length output from the fracture propagation model in the multi-level sensor network. ;

[0019] The reservoir productivity prediction model is used to predict production at different time points after fracturing. The input data includes reservoir thickness, porosity, permeability, fracturing section length collected by the multi-level sensor network, and fracture half-length output by the fracture propagation model. .

[0020] As an optional embodiment of the present invention, in a well site fracturing dynamic analysis method based on the Internet of Things, the generation of reservoir conductivity in step S3 includes: using a steady-state method, through formula... Calculate dimensionless conductivity Among them, crack permeability Based on the crack width acquired by the acoustic sensor in the multi-level sensor network Based on the fracturing fluid filtration coefficient, the reservoir matrix permeability is calculated. Reservoir logging data from the multi-level sensor network, fracture half-length Output by the crack propagation model;

[0021] The step S3 of generating the evaluation result of the fracturing equipment operating status includes: comparing the fracturing pump group pressure, flow rate, power and speed parameters collected in real time by the equipment layer sensors in the multi-level sensor network with the expected operating parameter range of the equipment obtained by simulation based on the digital twin model. If the real-time parameters continue to deviate from the expected range, the equipment operating status is determined to be abnormal.

[0022] As an optional embodiment of the present invention, in a well site fracturing dynamic analysis method based on the Internet of Things, step S4 includes: generating graded control instructions based on comprehensive evaluation results, sending them to the fracturing execution system through the Internet of Things control terminal, dynamically adjusting the fracturing parameters, and generating feedback data after control.

[0023] The tiered control instructions include primary control, secondary control, and tertiary control.

[0024] The first-level control is used to adjust the fracturing fluid discharge rate within a first preset range;

[0025] The secondary control is used to adjust the proppant concentration and fracturing fluid viscosity within a second preset range;

[0026] The three-level control is used to suspend the construction of the current fracturing section and start the well washing procedure, wherein the flow rate of the well washing fluid is controlled according to a preset ratio based on the current fracturing fluid discharge rate;

[0027] The IoT-based well site fracturing dynamic analysis method includes:

[0028] Step S5: Collect feedback data after adjustment in real time and compare it with the preset threshold. If the feedback data exceeds the threshold, repeat steps S3 to S4 until the feedback data is within the threshold range.

[0029] As an optional embodiment of the present invention, a well site fracturing dynamic analysis method based on the Internet of Things includes:

[0030] Step S6: The cloud-based analysis platform constructs a machine learning prediction model based on historical data to predict the output at different time scales after fracturing;

[0031] The training process of the machine learning prediction model includes:

[0032] Step S601: Collect fracturing operation parameters, reservoir parameters and corresponding production data of historical well sites to construct a training dataset;

[0033] Step S602: Standardize the training dataset;

[0034] Step S603: Divide the standardized dataset into a training set and a validation set according to a preset ratio, and use the gradient descent optimization algorithm to train the model. During the training process, set the initial learning rate and adaptively decay it according to a preset strategy.

[0035] Step S604: Calculate the prediction error of the model using the validation set. When the prediction error is lower than a preset threshold, stop training and save the final model parameters.

[0036] As an optional embodiment of the present invention, in a well site fracturing dynamic analysis method based on the Internet of Things, the multi-level sensor network in step S1 is divided into equipment layer sensors, reservoir sensors and environmental layer sensors according to the monitoring dimensions. The equipment layer sensors include pressure sensors, flow sensors, power sensors and speed sensors of the fracturing pump group. The reservoir sensors include distributed fiber optic strain sensors, downhole microseismic detectors and acoustic sensors. The environmental layer sensors include soil moisture sensors, groundwater quality sensors and air particulate matter sensors.

[0037] The fracturing fluid performance parameters in step S1 include viscosity, pH value, proppant concentration, and surface tension, which are collected by an online monitoring instrument installed in the fracturing fluid circulation pipeline.

[0038] This invention also provides an IoT-based well site fracturing dynamic analysis system, comprising:

[0039] Multi-level sensor network: Built at the well site, used to collect real-time operating parameters of fracturing equipment, dynamic response data of downhole reservoirs, performance parameters of fracturing fluid, and monitoring data of the surrounding environment of the well site, forming multi-source data;

[0040] The edge computing module is communicatively connected to the multi-level sensor network and is used to preprocess the collected multi-source data to obtain preprocessed data.

[0041] A cloud-based analysis platform communicates with the edge computing module. The platform includes a digital twin model building unit and a multi-dimensional analysis unit. The digital twin model building unit constructs a well site fracturing digital twin model. The multi-dimensional analysis unit calls a preset multi-dimensional analysis model to perform correlation calculations between the preprocessed data and the digital twin model, generating a comprehensive evaluation result of fracture propagation dynamic characteristics, reservoir conductivity, and fracturing equipment operating status. The multi-dimensional analysis model includes a fracture propagation model, a proppant migration model, and a reservoir productivity prediction model. The fracture propagation model is based on fracture mechanics theory and is used to calculate the dynamic characteristics of fracture propagation. The proppant migration model calculates the proppant migration law in the fracture based on the proppant concentration and fracturing fluid flow rate in the multi-source data. The reservoir productivity prediction model predicts the post-fracturing productivity based on reservoir parameters in the multi-source data.

[0042] The Internet of Things (IoT) control terminal communicates with the cloud-based analysis platform and is used to generate control commands based on the comprehensive evaluation results to dynamically adjust the fracturing parameters.

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0044] 1. In the IoT-based dynamic analysis method for well site fracturing of the present invention, multi-dimensional data is collected by constructing a multi-level sensor network, which is then preprocessed by edge computing and uploaded to a cloud analysis platform. Combined with a digital twin model, a comprehensive evaluation result is generated and a closed-loop feedback of hierarchical control is realized. This enables dynamic perception and precise control of the entire well site fracturing process, thereby improving the problem that traditional well site fracturing dynamic analysis mostly relies on local data monitoring and manual experience control. Due to incomplete data coverage and lack of real-time feedback, the fracturing effect deviates significantly from the design expectations.

[0045] 2. In the IoT-based dynamic analysis method for well site fracturing of the present invention, a digital twin model of well site fracturing is constructed on a cloud analysis platform. Based on real-time acquired data and physical field simulation model, correlation calculation is performed to achieve dynamic alignment between virtual simulation of fracturing process and actual monitoring. This improves the problem of low matching degree between numerical simulation model and actual reservoir conditions in traditional fracturing dynamic analysis, and insufficient accuracy of fracture propagation prediction due to difficulty in real-time calibration of model parameters.

[0046] 3. In the IoT-based well site fracturing dynamic analysis method of the present invention, a machine learning prediction model is constructed to predict the production at different times after fracturing. The model parameters are optimized by combining historical data and real-time evaluation results, thereby improving the long-term prediction accuracy of fracturing effect. This improves the problem that traditional production prediction after fracturing mostly relies on empirical formulas and does not fully associate with real-time fracturing dynamic data, resulting in a large deviation between production prediction and actual production. Attached Figure Description

[0047] Figure 1 The flowchart shows a dynamic analysis method for well site fracturing based on the Internet of Things proposed in this invention.

[0048] Figure 2 This is a schematic diagram of the sensor architecture in the device layer of the well site fracturing dynamic analysis method based on the Internet of Things proposed in this invention;

[0049] Figure 3 This is a schematic diagram of the reservoir sensor architecture in the IoT-based dynamic analysis method for well site fracturing proposed in this invention;

[0050] Figure 4 This is a schematic diagram of the environmental layer sensor architecture for a well site fracturing dynamic analysis method based on the Internet of Things proposed in this invention;

[0051] Figure 5 This is a schematic diagram of the architecture of a multi-dimensional analysis model for a well site fracturing dynamic analysis method based on the Internet of Things proposed in this invention;

[0052] Figure 6This is a schematic diagram of the data preprocessing steps in a well site fracturing dynamic analysis method based on the Internet of Things proposed in this invention;

[0053] Figure 7 This is a schematic diagram illustrating the training process of the machine learning prediction model in the IoT-based dynamic analysis method for well site fracturing proposed in this invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0055] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0056] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.

[0057] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0058] In the description of this invention, it should be noted that the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. These terms are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0059] like Figure 1 As shown, this embodiment of the invention provides a dynamic analysis method for well site fracturing based on the Internet of Things, including the following steps:

[0060] Step S1: Construct a multi-level sensor network at the well site. The multi-level sensor network is divided into equipment layer, reservoir layer and environment layer according to the monitoring dimensions. The multi-level sensor network is used to collect fracturing equipment operating parameters, downhole reservoir dynamic response data, fracturing fluid performance parameters and well site surrounding environment monitoring data in real time to form multi-source data.

[0061] Step S2: The collected multi-source data is preprocessed through edge computing nodes. The preprocessed data is then uploaded to the cloud analysis platform via an encrypted transmission protocol. The cloud analysis platform has a digital twin model of well site fracturing.

[0062] Step S3: The cloud-based analysis platform uses a multi-dimensional analysis model to perform correlation calculations on the preprocessed data and the digital twin model, generating a comprehensive evaluation result of the dynamic characteristics of fracture propagation, reservoir conductivity, and fracturing equipment operation status.

[0063] Step S4: Generate graded control instructions based on the comprehensive evaluation results, send them to the fracturing execution system through the Internet of Things control terminal, dynamically adjust the fracturing parameters, and generate feedback data after control.

[0064] Step S5: Collect feedback data after adjustment in real time and compare it with the preset threshold. If the feedback data exceeds the threshold, repeat steps S3 to S4 until the feedback data is within the threshold range.

[0065] This invention discloses an IoT-based dynamic analysis method for well site fracturing. By employing the aforementioned technical solution, a multi-level sensor network is constructed to collect multi-dimensional data. After edge computing preprocessing, the data is uploaded to a cloud analysis platform. Combined with a digital twin model, a comprehensive evaluation result is generated, and a closed-loop feedback for hierarchical control is achieved. This enables dynamic perception and precise control of the entire well site fracturing process, thereby improving upon the problem that traditional well site fracturing dynamic analysis mostly relies on local data monitoring and manual experience control. Due to incomplete data coverage and lack of real-time feedback, the fracturing effect deviates significantly from the design expectations.

[0066] Specifically, a multi-level sensor network is constructed at the well site, divided into equipment, reservoir, and environmental layers according to monitoring dimensions. This network is used to collect real-time operating parameters of fracturing equipment, dynamic response data of downhole reservoirs, performance parameters of fracturing fluid, and monitoring data of the surrounding environment of the well site to form multi-source data. By collecting multi-dimensional data in a hierarchical and classified manner, comprehensive, accurate, and original information input covering equipment, reservoir, and environmental levels is provided for subsequent IoT-based dynamic analysis of fracturing at the well site. This ensures that subsequent data preprocessing, correlation calculation, and control have a complete data foundation, guaranteeing the comprehensiveness and accuracy of fracturing dynamic analysis from the source of multi-dimensional data collection, and providing data support for the entire fracturing analysis process.

[0067] By performing preprocessing operations on the collected multi-source data through edge computing nodes, and then uploading the preprocessed data to a cloud-based analysis platform with a well site fracturing digital twin model via an encrypted transmission protocol, the edge computing nodes can perform preliminary processing of multi-source data, ensuring data quality and security. The encrypted transmission protocol ensures the confidentiality and integrity of the data transmission process. Based on the constructed well site fracturing digital twin model, the cloud-based analysis platform can conduct subsequent correlation calculations and analyses based on valid data, providing reliable data processing, transmission, and model support for fracturing dynamic analysis.

[0068] The cloud-based analytics platform utilizes a multi-dimensional analysis model to correlate pre-processed multi-source data with the well site fracturing digital twin model, generating a comprehensive assessment of fracture propagation dynamics, reservoir conductivity, and fracturing equipment operational status. This allows for the integration of information from pre-processed data and the digital twin model, enabling a systematic analysis of key fracturing operation indicators and outputting comprehensive evaluation results. This provides precise, multi-dimensional decision-making support for subsequent fracturing parameter control, facilitating seamless integration of critical stages in the fracturing dynamic analysis process, from data processing to control command generation.

[0069] Based on comprehensive evaluation results, tiered control commands are generated and transmitted to the fracturing execution system via IoT control terminals. This dynamically adjusts fracturing parameters and generates feedback data after control, allowing for precise intervention in fracturing operations based on multi-dimensional analysis. The IoT terminal enables command transmission and execution, dynamically optimizing fracturing parameters while collecting feedback data to provide real-time support for subsequent analysis and potential further control. This forms a closed loop from evaluation to control to feedback, ensuring fracturing operations are adjusted as needed and run efficiently, seamlessly connecting the application of analysis results with the optimization of construction parameters in the dynamic fracturing analysis process.

[0070] Real-time feedback data after adjustment is collected and compared with preset thresholds. If the feedback data exceeds the threshold, steps S3 to S4 are repeated until the feedback data falls within the threshold range. This allows for real-time monitoring of the adjustment effect and threshold comparison to determine whether the fracturing parameter adjustment has achieved the expected results. When the expected results are not achieved, a cyclical mechanism of re-analysis and adjustment is initiated to ensure that the fracturing operation status remains within a reasonable range, forming a complete closed loop from feedback monitoring to dynamic optimization. This ensures the continuity and effectiveness of fracturing dynamic analysis and connects the fracturing parameter adjustment with the stability of the operation status.

[0071] By constructing a multi-level sensor network to collect multi-dimensional data, which is then preprocessed using edge computing and uploaded to a cloud-based analysis platform, a comprehensive evaluation result is generated in conjunction with a digital twin model. This enables closed-loop feedback for tiered control, thereby achieving dynamic perception and precise control of the entire well site fracturing process. This improves upon traditional well site fracturing dynamic analysis, which largely relies on localized data monitoring and manual experience-based control. Due to incomplete data coverage and a lack of real-time feedback, this often results in significant deviations between fracturing effects and design expectations.

[0072] like Figures 2-5 As shown, in step S1, the equipment layer sensors include pressure sensors, flow sensors, power sensors, and speed sensors of the fracturing pump group; the reservoir sensors include distributed fiber optic strain sensors, downhole microseismic detectors, and acoustic sensors; and the environmental layer sensors include soil moisture sensors, groundwater quality sensors, and air particulate matter sensors. The fracturing fluid performance parameters include viscosity, pH value, proppant concentration, and surface tension, which are collected by an online monitoring instrument installed in the fracturing fluid circulation pipeline.

[0073] Specifically, pressure, flow, power, and speed sensors in the equipment layer are used to collect operating parameters of the fracturing pump unit; distributed fiber optic strain sensors, downhole microseismic detectors, and acoustic sensors in the reservoir layer are used to collect dynamic response data of the downhole reservoir; soil moisture sensors, groundwater quality sensors, and air particulate matter sensors in the environmental layer are used to collect monitoring data of the surrounding environment of the well site; and online monitoring instruments installed in the fracturing fluid circulation pipeline are used to collect performance parameters of the fracturing fluid such as viscosity, pH value, proppant concentration, and surface tension. This configuration of specific sensors and online monitoring instruments at each level enables targeted data collection on the operating status of fracturing equipment, downhole reservoir response, surrounding environment of the well site, and fracturing fluid performance. This ensures the comprehensiveness and accuracy of multi-source data acquired by the multi-level sensor network across equipment, reservoir, environment, and fracturing fluid dimensions, providing specific and suitable raw data support for subsequent data preprocessing and analysis.

[0074] like Figure 6 As shown, in step S2, data preprocessing includes the following steps:

[0075] Step S201: Use the Laida criterion to remove outliers from the data, using the formula... Identify outliers, among which For a single sensor data point, This represents the average value of data within the same time period. The standard deviation is used to identify outliers and remove them.

[0076] Step S202: Align the data after removing outliers with the time axis, and use the start-up time of the fracturing pump as the reference zero point to convert the acquisition timestamps of different sensors into relative times.

[0077] Step S203: Eliminate the influence of dimensions by normalization, using the formula... Normalization was performed, where For the normalized data, The minimum value of the data. The maximum value of the data;

[0078] The encrypted transmission protocol uses the AES-256 encryption algorithm, combined with Algorithm pair The key is used for encryption, and the encrypted data packet contains the data acquisition timestamp, sensor number, and CRC32 checksum.

[0079] Specifically, in step S2, the Raida criterion is used through the formula... Remove outliers, among which, The data for a single sensor line comes from multiple sources, including sensors at the equipment layer, reservoir environment layer, and online fracturing fluid monitoring instrument. The average value of data for the same time period is calculated by analyzing the data collected during the same period. The arithmetic mean is obtained. The standard deviation is calculated based on data from the same time period. This formula takes the standard deviation of the average value of a single sensor data point within the same time period as input and outputs the outlier determination result; data identified as outliers are removed. The data after outlier removal is then time-axis aligned, using the fracturing pump start-up time as the baseline zero point. The acquisition timestamps from different sensors are uniformly converted to relative times, ensuring synchronization of multi-source data in the time dimension. Normalization is then performed using the formula... Eliminating the influence of dimensions, among which, For the normalized data, This is a single data entry after outlier removal and time alignment. The minimum value among the same type of data. The maximum value among the same type of data. and By statistically analyzing similar data from the preprocessing stage, this formula takes the processed single data point and the maximum / minimum values ​​of similar data as input, and outputs normalized data. The data processed by the above formula is uploaded to the cloud analysis platform via an encrypted transmission protocol for correlation calculation with the well site fracturing digital twin model. This provides standardized input data for the multi-dimensional analysis model, ensuring the accuracy of subsequent comprehensive evaluation results. The AES key is encrypted using an AES-256 encryption algorithm combined with an RSA algorithm, and the encrypted data packet includes the data acquisition timestamp, sensor number, and CRC32 checksum. This allows for the removal of outliers, ensuring data accuracy. Time axis alignment achieves time synchronization of multi-source data, normalization eliminates differences in data dimensions, and the encrypted transmission protocol ensures data transmission security and traceability. This ensures that the preprocessed data meets the requirements of the cloud analysis platform for data quality, consistency, and security in subsequent correlation calculations, providing reliable processing and transmission guarantees for the data flow from acquisition to analysis.

[0080] In step S2, the construction of the well site fracturing digital twin model includes: based on the well site three-dimensional geological modeling data, well completion parameters and historical fracturing data, a physical field simulation model is constructed using finite element analysis software. The physical field includes stress field, seepage field and temperature field. The update frequency of the digital twin model is consistent with the sensor data acquisition frequency.

[0081] Specifically, in step S2, based on the well site's 3D geological modeling data, well completion parameters, and historical fracturing data, a physical field simulation model including stress field, seepage field, and temperature field is constructed using finite element analysis software to form a digital twin model of the well site fracturing. The update frequency of this model is consistent with the sensor data acquisition frequency. Thus, a realistic physical field simulation model can be constructed using 3D geological, well completion, and historical fracturing data. By covering stress field, seepage field, and temperature field, a comprehensive simulation of the physical processes related to well site fracturing is achieved. The update frequency, consistent with the sensor data acquisition frequency, ensures the synchronization between the model and the real-time acquired data, enabling the digital twin model to provide a dynamic and realistic virtual mapping basis for the associated calculations of the cloud analysis platform.

[0082] like Figure 5 As shown, in step S3, the multi-dimensional analysis model includes a fracture propagation model, a proppant migration model, and a reservoir productivity prediction model; the fracture propagation model is constructed based on fracture mechanics theory, and is expressed through formulas... Calculate the type I fracture toughness, where, Type I fracture toughness, Geometric factor For reservoir geostress, The crack half-length is used as the input data, which consists of strain values ​​collected by distributed fiber optic strain sensors, source location data collected by microseismic detectors, and geostress logging data.

[0083] Specifically, in step S3, the multi-dimensional analysis model includes a fracture propagation model, a proppant migration model, and a reservoir productivity prediction model, which are calculated using formulas. Calculate the type I fracture toughness, where Type I fracture toughness is a key indicator for assessing crack propagation capacity. The geometric factor is determined by the fracture morphology and boundary conditions based on the well site 3D geological modeling data using finite element analysis software. This represents reservoir in-situ stress, derived from in-situ stress logging data. The fracture half-length is obtained by inverting strain values ​​acquired by distributed fiber optic strain sensors and source location data acquired by microseismic detectors. The formula takes geometric factors, reservoir stress, and fracture half-length as inputs and outputs Type I fracture toughness. This result is used to determine whether fracture propagation meets expectations. If the fracture toughness threshold is exceeded, the fracture will continue to propagate; otherwise, it will tend to stabilize. The output results serve as input parameters for the proppant migration model, used to calculate the proppant migration pattern in the fractures. They also participate in the calculation of reservoir conductivity, providing a basis for staged control commands. When fracture propagation does not meet expectations, the system triggers adjustments to the fracturing parameters to ensure that the fracturing effect meets design requirements.

[0084] The proppant transport model is based on the formula. Describe the proppant transport behavior, among which, For proppant concentration, For time, This refers to the fracturing fluid flow rate. The coordinates are along the crack length direction. The diffusion coefficient is... The proppant settling coefficient is used; the input data includes fracturing fluid flow sensor data, proppant concentration detector data, and fracture half-length output from the fracture propagation model. .

[0085] Specifically, the proppant transport model is derived through the formula... Describe the proppant transport behavior, among which, The proppant concentration is obtained by real-time data collection from the fracturing fluid circulation pipeline using a proppant concentration detector. The time is converted to relative time by aligning with the time axis, with the start-up time of the fracturing pump as the reference zero point; The fracturing fluid velocity is calculated from data from the fracturing fluid flow sensor. The coordinates are the crack length direction coordinates, and the crack half-length is output based on the crack propagation model. Determine spatial location; The diffusion coefficient is pre-calibrated based on historical fracturing data and reservoir characteristics; The proppant settling coefficient, measured through laboratory simulation experiments, is used as input to calculate the spatiotemporal distribution of proppant in the fracture using partial differential equations. The output is the proppant concentration distribution at different times and locations within the fracture. This result is used to generate the proppant transport dynamic characteristics in the comprehensive evaluation results, providing a basis for tiered control commands. When the proppant concentration distribution does not meet expectations, the system will trigger primary control to fine-tune the fracturing fluid discharge rate, or initiate secondary control to adjust the proppant concentration and fracturing fluid viscosity, ensuring a reasonable distribution of proppant in the fracture to improve fracturing efficiency.

[0086] In step S4, the graded control instructions include primary control, secondary control, and tertiary control. Primary control is used to fine-tune the fracturing fluid discharge rate, with an adjustment range of 5% to 10% of the current discharge rate. Secondary control includes adjusting the proppant concentration and fracturing fluid viscosity. The proppant concentration adjustment range is 0.1 kg / L to 0.3 kg / L, and the viscosity adjustment is achieved by adding a thickener, with a viscosity adjustment range of 5 mPa·s to 20 mPa·s. Tertiary control is used to suspend the corresponding fracturing section construction and start the well washing procedure. After starting the well washing procedure, the well washing fluid will be discharged, and the well washing fluid flow rate is 1.2 to 1.5 times the fracturing fluid discharge rate.

[0087] Specifically, in step S4, the graded control instructions include: Level 1 control for fine-tuning the fracturing fluid discharge rate, with an adjustment range of 5% to 10% of the current discharge rate; Level 2 control for adjusting the proppant concentration, with an adjustment range of 0.1 kg / L to 0.3 kg / L, and adjusting the fracturing fluid viscosity by adding a thickener, with an adjustment range of 5 mPa·s to 20 mPa·s; and Level 3 control for pausing the corresponding fracturing section and initiating the well-washing procedure, with the flow rate of the well-washing fluid discharged after initiation being 1.2 to 1.5 times the fracturing fluid discharge rate. Thus, Level 1 control enables small-scale precise adjustments to the fracturing fluid discharge rate, Level 2 control enables targeted adjustments to the proppant concentration and fracturing fluid viscosity, and Level 3 control enables pausing the operation when necessary and discharging the well-washing fluid at a specific flow rate. This forms a graded response mechanism from minor adjustments to emergency treatment, enabling the fracturing execution system to dynamically adjust the fracturing parameters according to the comprehensive evaluation results, ensuring that the fracturing operation proceeds as expected.

[0088] This embodiment of the well site fracturing dynamic analysis method based on the Internet of Things further includes S6: A cloud analysis platform constructs a machine learning prediction model based on historical data to predict production at 30, 90, and 180 days after fracturing; the machine learning prediction model uses a long short-term memory network, and the formula is used to predict the production at 30, 90, and 180 days after fracturing. Calculate the loss value, where For actual output, To predict production, For the sample size, The regularization coefficient is . These are the model weight parameters. The number of weighted parameters; input data includes reservoir thickness, porosity, permeability, fracturing section length, and fracture half-length from the comprehensive evaluation results. .

[0089] Specifically, in step S6, the machine learning prediction model built by the cloud analysis platform based on historical data uses a long short-term memory network, and is implemented through the formula... Calculate the loss value, where, The actual production data comes from the production statistics of the corresponding time period of the historical well site. The predicted production figures are the results of machine learning prediction models for production at 30, 90, and 180 days after fracturing. The sample size is determined by the total number of historical data samples collected. The regularization coefficients are pre-set according to the model training requirements. These are the model weight parameters, which are automatically learned during model training. The weight parameters are determined by the network structure of the machine learning prediction model. This formula takes the actual production and predicted production sample size, regularization coefficient, model weight parameters, and the number of weight parameters as input, and outputs a loss value to measure the prediction error and weight parameter complexity of the machine learning prediction model. This result is used in the training process of the machine learning prediction model. The Adam optimizer adjusts the model weight parameters to reduce the loss value. When the prediction error corresponding to the loss value is less than 5%, training stops and the model parameters are saved. The trained model can predict post-fracturing production based on input data such as reservoir thickness, porosity, permeability, fracturing section length, and fracture half-length 'a', providing a basis for long-term evaluation of fracturing effectiveness. Thus, by leveraging long short-term memory networks and machine learning prediction models with specific loss functions, combined with reservoir parameters such as fracturing section length and fracture half-length, it is possible to predict production at different times after fracturing, providing forward-looking information for long-term evaluation of fracturing effectiveness and supporting the judgment of the effectiveness of fracturing schemes.

[0090] like Figure 7 As shown, in step S6, the training process of the machine learning prediction model includes:

[0091] Collect fracturing operation parameters, reservoir parameters, and production data for the corresponding time periods from historical well sites to construct a training dataset;

[0092] The training dataset is standardized using the formula. Standardization processing is performed, among which This is the original data. The mean of the dataset. The standard deviation of the dataset;

[0093] The standardized dataset is divided into The model was divided into training and validation sets. The Adam optimizer was used to train the model. The initial learning rate was set to 0.001, and the learning rate was reduced to 0.9 times the previous value every 100 iterations.

[0094] The model prediction error is calculated using the validation set. When the error is less than... Stop training and save the model parameters.

[0095] Specifically, in step S6, a training dataset is constructed by collecting fracturing operation parameters, reservoir parameters, and production data for corresponding time periods from historical well sites. This dataset is then processed using the formula... The training dataset is standardized, whereby... The raw data includes fracturing operation parameters, reservoir parameters, and production data for the corresponding time periods from historical well sites. This data was obtained by collecting and organizing historical records. The dataset mean is the arithmetic mean calculated from all the original data in the training dataset. The standard deviation of a dataset is calculated based on the deviation of all original data points in the training dataset from the mean. This formula takes the original dataset mean and the dataset standard deviation as input and outputs the standardized dataset. This result is used to convert raw data of different dimensions into data of a uniform scale. The next step is to divide the standardized dataset into training and validation sets in a 7:3 ratio. The training set is used for parameter learning of the machine learning prediction model, and the validation set is used to calculate the model's prediction error. Training is stopped and the model parameters are saved when the error is less than 5%. Standardization ensures a balanced weighting of the influence of each feature on the prediction result during model training, improving the training efficiency and prediction accuracy of the machine learning prediction model and providing a reliable model foundation for post-fracturing yield prediction. The standardized dataset is divided into training and validation sets in a 7:3 ratio. The Adam optimizer is used to train the model, with an initial learning rate of 0.001, which decays to 0.9 times the previous rate every 100 iterations. The prediction error is calculated using the validation set. Training is stopped and the model parameters are saved when the error is less than 5%. This allows for the construction of a training dataset containing historical parameters and yield data, followed by standardization. Combined with a reasonably divided training and validation set, and the set optimizer parameters and stopping conditions, the machine learning prediction model can reach a preset accuracy after training, ensuring the model has reliable yield prediction capabilities and providing model performance assurance for post-fracturing yield prediction.

[0096] This invention provides an IoT-based dynamic analysis method for well site fracturing, which calculates reservoir conductivity using a steady-state method and a formula. Calculate the dimensionless conductivity, where For dimensionless flow conduction capability, For crack permeability, The width of the crack. For reservoir matrix permeability, Fracture half-length: Input data includes fracture width data collected by acoustic sensors, reservoir logging permeability data, and fracture half-length output from the fracture propagation model. .

[0097] Specifically, the calculation of reservoir conductivity in step S3 adopts the steady-state method, using the formula... Calculate the dimensionless conductivity, where Dimensionless conductivity is a key indicator for measuring reservoir conductivity efficiency. The fracture permeability is calculated by combining the fracture width data collected by the acoustic sensor with the fracturing fluid filtration coefficient. The crack width is directly derived from real-time data collected by the acoustic sensor. The reservoir matrix permeability is obtained by standardizing reservoir logging permeability data. The fracture half-length is calculated and output by the fracture propagation model based on strain values ​​acquired by distributed fiber optic strain sensors, source location data acquired by microseismic detectors, and geostress logging data. This formula takes fracture permeability, fracture width, reservoir matrix permeability, and fracture half-length as inputs and outputs dimensionless conductivity. This result is used to evaluate the conductivity of the reservoir after fracturing. If the value is greater than the design threshold, it indicates that the fracture conductivity is good and the proppant distribution is reasonable; otherwise, a graded control command needs to be triggered to optimize the conductivity by adjusting the fracturing fluid discharge, proppant concentration or starting the well washing procedure, so as to ensure that the oil and gas can flow into the wellbore efficiently after fracturing.

[0098] A dynamic analysis system for well site fracturing based on the Internet of Things, comprising:

[0099] A multi-level sensor network is used to collect fracturing equipment operating parameters, downhole reservoir dynamic response data, fracturing fluid performance parameters, and well site surrounding environment monitoring data in real time according to monitoring dimensions.

[0100] The edge computing module communicates with a multi-level sensor network and is used to preprocess the collected multi-source data;

[0101] The cloud-based analysis platform communicates with the edge computing module. Internally, it includes a digital twin model building unit and a multi-dimensional analysis unit. The digital twin model building unit constructs a digital twin model of the well site fracturing, while the multi-dimensional analysis unit calls a pre-defined multi-dimensional analysis model to perform correlation calculations between pre-processed data and the digital twin model, generating a comprehensive evaluation result of fracture propagation dynamics, reservoir conductivity, and fracturing equipment operating status. The multi-dimensional analysis model includes a fracture propagation model, a proppant migration model, and a reservoir productivity prediction model. The fracture propagation model, based on fracture mechanics theory, calculates the dynamic characteristics of fracture propagation. The proppant migration model calculates the proppant migration pattern in the fracture based on the proppant concentration and fracturing fluid velocity in the multi-source data. The reservoir productivity prediction model predicts the post-fracturing productivity based on reservoir parameters in the multi-source data.

[0102] The Internet of Things (IoT) control terminal communicates with the cloud-based analysis platform to generate tiered control commands based on comprehensive evaluation results, and then sends these commands to the fracturing execution system, which dynamically adjusts the fracturing parameters.

[0103] The cloud-based analytics platform is also used to receive feedback data after regulation. When the feedback data exceeds the preset threshold, it triggers the multi-dimensional analysis unit to recalculate the correlation and issues new hierarchical regulation instructions through the IoT control terminal until the feedback data is within the threshold range.

[0104] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.

Claims

1. An Internet of Things based wellsite fracturing dynamic analysis method, characterized in that, The method comprises the following steps: Step S1: A multi-level sensor network is constructed at the well site for real-time acquisition of fracturing equipment operation parameters, downhole reservoir dynamic response data, fracturing fluid performance parameters and well site surrounding environment monitoring data, forming multi-source data; Step S2: The collected multi-source data is preprocessed by an edge computing node to obtain preprocessed data, and the preprocessed data is uploaded to a cloud analysis platform through an encryption transmission protocol, and the cloud analysis platform constructs a well site pressure digital twin model; Step S3: The cloud analysis platform performs associated calculation on the preprocessed data and the digital twin model through a multi-dimensional analysis model to generate a comprehensive evaluation result of fracture expansion dynamic characteristics, reservoir conductivity and fracturing equipment operation state, the multi-dimensional analysis model comprising a fracture expansion model, a proppant migration model and a reservoir productivity prediction model, the fracture expansion model being constructed based on fracture mechanics theory for calculating fracture expansion dynamic characteristics, the proppant migration model being based on proppant concentration and fracturing fluid flow rate in the multi-source data to calculate proppant migration law in the fracture, and the reservoir productivity prediction model being based on reservoir parameters in the multi-source data to predict productivity after fracturing; Step S4: Based on the comprehensive evaluation result, a regulation and control instruction is generated for dynamic adjustment of fracturing parameters. The generation of reservoir conductivity in step S3 includes: using a steady-state method, through a formula... Calculate dimensionless conductivity Among them, crack permeability Based on the crack width acquired by the acoustic sensor in the multi-level sensor network Based on the fracturing fluid filtration coefficient, the reservoir matrix permeability is calculated. Reservoir logging data from the multi-level sensor network, fracture half-length Output by the crack propagation model; In the step S3, the evaluation result of the fracturing equipment operation state is generated by comparing the real-time parameters of the fracturing pump pressure, flow, power and rotating speed collected by the device layer sensor in the multi-level sensor network with the expected operation parameter interval obtained based on the digital twin model simulation, and if the real-time parameters continuously deviate from the expected interval, it is determined that the equipment operation state is abnormal.

2. The method of claim 1, wherein, In the step S2, the multi-source data after removing abnormal values is time axis aligned with the fracturing pump start time as the reference zero point, and the collection time stamps of different sensors are uniformly converted into relative time. Step S201: determine the abnormal value in multi-source data by formula determine the abnormal value in multi-source data by formula for single sensor data, for the average value of data in the same period, for the standard deviation, the data satisfying the condition is determined as an abnormal value, and the abnormal value is removed by using the Laplace criterion. In the step S2, the cloud analysis platform constructs a well site pressure digital twin model, which comprises: based on well site three-dimensional geological modeling data, completion parameters and historical fracturing data, a physical field simulation model including stress field, seepage field and temperature field is constructed by a finite element analysis software to form a well site pressure digital twin model, and the update frequency of the well site pressure digital twin model is consistent with the sensor data acquisition frequency of the multi-level sensor network. The step S4 comprises: based on the comprehensive evaluation result, a hierarchical regulation and control instruction is generated, which is issued to the fracturing execution system through an Internet of Things control terminal for dynamic adjustment of fracturing parameters and generation of feedback data after regulation and control; 3. The method of claim 1, wherein, The hierarchical regulation and control instruction comprises primary regulation and control, secondary regulation and control and tertiary regulation and control; 4. The method of claim 1, wherein, The crack propagation model is calculated by the formula The mode I fracture toughness is calculated, wherein KIC is the mode I fracture toughness, A is a geometric factor, σh is the in-situ stress of the reservoir, L is the half-length of the fracture, and the input data includes strain values collected by distributed optical fiber strain sensors in the multi-level sensor network, source positioning data collected by microseismic detectors, and geostress logging data; The proppant transport model is given by the formula The proppant transport model is given by the formula is the proppant concentration, is time, is the fracturing fluid flow rate, and is the fracture length direction coordinate, is the diffusion coefficient, is the proppant settling coefficient, and the input data includes fracturing fluid flow sensor data in the multi-level sensor network, proppant concentration detector data, and the fracture half-length output by the fracture propagation model ; The reservoir productivity prediction model is used to predict the production at different time periods after fracturing, and the input data includes the reservoir thickness, porosity, permeability, fracturing segment length collected by the multi-level sensor network and the fracture half-length output by the fracture propagation model .

5. The method of claim 1, wherein, The primary regulation and control is used for adjusting the fracturing fluid displacement within a first preset amplitude range; The secondary regulation and control is used for adjusting the proppant concentration and the fracturing fluid viscosity within a second preset amplitude range; ​ ​ The three-level regulation is used to suspend the construction of the current fracturing section and start a well flushing program, wherein the flow rate of the well flushing fluid is controlled according to a preset ratio based on the current fracturing fluid discharge; The well site fracturing dynamic analysis method based on the Internet of Things comprises: Step S5: Real-time acquisition of the feedback data after regulation is compared with a preset threshold value, if the feedback data exceeds the threshold value, steps S3 to S4 are repeated until the feedback data is within the threshold value range.

6. The method of claim 1, wherein, The training process of the reservoir productivity prediction model comprises: Step S601: Collecting fracturing construction parameters, reservoir parameters and corresponding production data of historical well sites to construct a training data set; Step S602: Standardizing the training data set; Step S603: Dividing the standardized data set into a training set and a validation set according to a preset proportion, and training the model using a gradient descent optimization algorithm, setting an initial learning rate and performing adaptive decay according to a preset strategy during the training process; Step S604: Calculating the prediction error of the model through the validation set, stopping training when the prediction error is lower than a preset threshold value, and saving the final model parameters.

7. The method of claim 1, wherein, The multi-level sensor network in step S1 is divided into device layer sensors, reservoir sensors and environment layer sensors according to monitoring dimensions, the device layer sensors comprise pressure sensors, flow sensors, power sensors and speed sensors of the fracturing pump group, the reservoir sensors comprise distributed optical fiber strain sensors, downhole microseismic detectors and acoustic sensors, and the environment layer sensors comprise soil moisture content sensors, underground water quality sensors and air particulate matter sensors; The fracturing fluid performance parameters in step S1 comprise viscosity, pH value, proppant concentration and surface tension, which are collected by an online detector arranged in the fracturing fluid circulation pipeline.

8. An Internet of Things based wellsite fracturing dynamic analysis system, characterized in that, Comprise: A multi-level sensor network is constructed at the well site for real-time acquisition of fracturing equipment operation parameters, downhole reservoir dynamic response data, fracturing fluid performance parameters and well site surrounding environment monitoring data to form multi-source data; An edge computing module is in communication connection with the multi-level sensor network and is used for pre-processing the collected multi-source data to obtain pre-processed data; A cloud analysis platform is in communication connection with the edge computing module, the cloud analysis platform is internally provided with a digital twin model construction unit and a multi-dimensional analysis unit, the digital twin model construction unit is used for constructing a well site fracturing digital twin model, and the multi-dimensional analysis unit is used for calling a preset multi-dimensional analysis model, performing associated calculation on the pre-processed data and the digital twin model, generating a comprehensive evaluation result of fracture expansion dynamic characteristics, reservoir conductivity and fracturing equipment operation state, the multi-dimensional analysis model comprises a fracture expansion model, a proppant transport model and a reservoir productivity prediction model, the fracture expansion model is constructed based on fracture mechanics theory and is used for calculating fracture expansion dynamic characteristics, the proppant transport model calculates the proppant transport law in the fracture based on the proppant concentration and the fracturing fluid flow rate in the multi-source data, and the reservoir productivity prediction model predicts the productivity after fracturing based on the reservoir parameters in the multi-source data; The Internet of Things control terminal is in communication connection with the cloud analysis platform and is configured to generate a regulation instruction based on the comprehensive evaluation result and dynamically adjust the fracturing parameters. The cloud-based analysis platform generates reservoir conductivity by employing a steady-state method and using formulas. Calculate dimensionless conductivity The crack permeability is based on the crack width collected by the acoustic sensors in the multi-level sensor network. Based on the fracturing fluid filtration coefficient, the reservoir matrix permeability is calculated. Reservoir logging data from the multi-level sensor network, fracture half-length Output by the crack propagation model; The cloud analysis platform generates an evaluation result of the operation state of the fracturing equipment, including: comparing the pressure, flow, power and rotating speed parameters of the fracturing pump group collected by the equipment layer sensors in the multi-level sensor network in real time with the expected operation parameter interval of the equipment obtained based on the digital twin model simulation, and if the real-time parameters continuously deviate from the expected interval, it is determined that the operation state of the equipment is abnormal.

Images