A cementing quality evaluation system and method for deep shale gas wells

Abstract

This invention discloses a cementing quality evaluation system and method for deep shale gas wells, belonging to the technical field of cementing quality evaluation for deep shale gas wells. The cementing quality evaluation system for deep shale gas wells includes a data collection component connected to a data set module for data aggregation. The data set module is connected to a strength evaluation component for assessing the strength of cement sheath bonding and a network training component for training a neural network. The network training component is connected to a structure determination component for determining the neural network structure. The structure determination component is connected to a real-time evaluation component for evaluating the cementing quality of shale gas wells. The network training component is connected to the real-time evaluation component. This invention effectively solves the problems of long evaluation time, high cost, and inaccurate evaluation results in existing shale gas well cementing quality evaluation methods.

Description

Technical Field

[0001] This invention relates to the field of cementing quality evaluation technology for deep shale gas wells, and specifically to a cementing quality evaluation system and method for deep shale gas wells. Background Technology

[0002] Shale gas is natural gas contained in shale formations. Due to the high density and low permeability of shale, conventional vertical well extraction would result in very low shale gas production. To increase production, horizontal well extraction is commonly used in shale gas reservoirs. For example, current shale gas reservoir development primarily utilizes cluster wells, each typically containing one horizontal well. The horizontal section is generally 1000m to 4286m long. Because of the long horizontal well sections, indiscriminate fracturing of shale gas horizontal wells is unlikely to achieve ideal fracturing results. Therefore, shale gas extraction usually requires segmented extraction. Fracturing is used to generate or connect more reservoir fractures to increase shale gas production. When performing staged fracturing on shale gas wells, the well must first be isolated in stages. The quality of cementing in shale gas wells directly determines the isolation and gas tightness of the stages in the horizontal well. In addition, the quality of cementing in shale gas wells plays a crucial role in the production, lifespan, and resource protection of shale gas wells. Therefore, cementing quality must be measured after cementing completion. However, because the horizontal section of a shale gas horizontal well is relatively long, conventional wireline logging methods can only measure the vertical section and part of the inclined section, and cannot measure the horizontal section.

[0003] Currently, existing methods for assessing cementing quality in horizontal sections include using a crawler to bring the logging instrument into the horizontal section. However, using a crawler is slow, and for the relatively long shale gas horizontal sections, assessing cementing quality typically takes several days. Furthermore, the crawler itself is expensive. Therefore, overall, there is currently no assessment technology for shale gas well cementing quality that is both time-efficient and cost-effective. Further, to improve the accuracy of the assessment results, more comprehensive and accurate data is needed. Simultaneously, the technology should be able to promptly identify and resolve cementing quality issues, improving assessment efficiency and accuracy, and contributing to a more comprehensive evaluation of cementing quality.

[0004] Patent CN111411937A discloses a cementing quality evaluation method and device. Specifically, it discloses obtaining cementing section length data through acoustic amplitude logging data, determining micro-well section length data based on cementing section length data, and then determining micro-well section length data based on well depth data and acoustic amplitude logging data. It uses cementing section length data and micro-well section length data to obtain the excellent rate, qualified rate and unqualified rate of cementing section. However, this patent has the problem of relatively complex evaluation. Summary of the Invention

[0005] The purpose of this invention is to provide a cementing quality evaluation system and method for deep shale gas wells, which solves the technical problems of long evaluation time, high cost and inaccurate evaluation results in existing shale gas well cementing quality evaluation methods.

[0006] To achieve the above objectives, one embodiment of the present invention provides a cementing quality evaluation system for deep shale gas wells, including a data collection component, a data set module connected to the data collection component, a strength evaluation component for evaluating the strength of cement sheath bonding and a network training component for training a neural network, a structure determination component for determining the structure of the neural network connected to the network training component, and a real-time evaluation component for evaluating the cementing quality of shale gas wells, wherein the network training component is connected to the real-time evaluation component.

[0007] In one preferred embodiment of the present invention, the data collection component includes a construction parameter collection module and a well logging data collection module. The well logging data collection module is connected to a well logging data processing module for processing well logging data. The well logging data processing module is connected to a well logging data fusion module for fusion of well logging data. The well logging data fusion module is connected to a well logging feature output module for extracting feature parameters. The well logging feature output module is connected to a data center processing module. The data center processing module is connected to the construction parameter collection module. The data center processing module is also connected to a construction parameter output module for outputting construction parameters and a well logging data output module for outputting well logging data. The construction parameter output module and the well logging data output module are connected to a data set module.

[0008] In one preferred embodiment of the present invention, the construction parameters collected by the construction parameter collection module include the cement slurry mix ratio, pumping pressure, pumping time, well depth, and formation characteristics.

[0009] In one preferred embodiment of the present invention, the strength assessment component includes a well logging data extraction module connected to a data set module, the well logging data extraction module being connected to a first interface assessment module and a second interface assessment module, the first interface assessment module being connected to a strength data output module, and the strength data output module being connected to the second interface assessment module.

[0010] In one preferred embodiment of the present invention, the network training component includes a construction parameter extraction module and a strength data extraction module. The strength data extraction module is connected to a data information processing module. The data information processing module is connected to a neural network selection module and a network training iteration module. The network training iteration module is connected to a network structure determination module for determining the neural network.

[0011] In one preferred embodiment of the present invention, the network training component includes a network performance evaluation module connected to the network structure determination module, and the network performance evaluation module is connected to a network storage and loading module for storing the neural network.

[0012] In one preferred embodiment of the present invention, the structure determination component includes a network training and analysis module, which is connected to a structure determination principle module, a layer node setting module, and a connection relationship determination module.

[0013] In one preferred embodiment of the present invention, the network training and analysis module is further connected to a weight threshold adjustment module for optimizing neural network performance and a performance stability detection module for detecting neural network performance.

[0014] In one preferred embodiment of the present invention, the real-time evaluation component includes a real-time network system module, which is connected to a neural network embedding module, and the neural network embedding module is connected to a real-time quality evaluation module for evaluating the cementing quality of deep shale gas wells.

[0015] In one preferred embodiment of the present invention, the real-time evaluation component includes a threshold alarm setting module connected to the real-time quality evaluation module.

[0016] In one preferred embodiment of the present invention, the real-time quality evaluation module is connected to a data interface display module for displaying results, and the data interface display module is connected to a result-assisted decision-making module.

[0017] This invention also discloses a cementing quality evaluation method for deep shale gas wells, based on the aforementioned cementing quality evaluation system for deep shale gas wells, comprising the following steps:

[0018] Acquire construction parameters and logging data for shale gas wells, and perform data processing on the construction parameters and logging data;

[0019] Based on well logging data, determine the bonding strength of the first interface and the bonding strength of the second interface;

[0020] Neural network training is performed based on construction parameters, the bonding strength of the first interface, and the bonding strength of the second interface.

[0021] Adjusting training parameters during neural network training to determine the neural network structure;

[0022] Cementing quality is evaluated based on the determined neural network structure.

[0023] One preferred embodiment of the present invention involves obtaining construction parameters and logging data of a shale gas well, and processing the construction parameters and logging data, including:

[0024] Obtain construction parameters and logging data for shale gas wells;

[0025] The acquired well logging data is preprocessed, fused, and its characteristic parameters are extracted.

[0026] The construction parameters are matched with the well logging data after feature extraction.

[0027] One preferred embodiment of the present invention involves training a neural network based on construction parameters, the bonding strength of the first interface, and the bonding strength of the second interface, including:

[0028] Neural network selection based on data features of construction parameters;

[0029] The selected neural network is trained based on construction parameters, the bonding strength of the first interface, and the bonding strength of the second interface.

[0030] In one preferred embodiment of the present invention, the performance of the selected neural network is evaluated during the training process.

[0031] One preferred embodiment of the present invention is to regulate the training parameters in neural network training to determine the neural network structure, including: regulating the weights, thresholds, input layer, output layer and hidden layer of the neural network to determine the neural network structure.

[0032] In summary, the beneficial effects of the present invention are as follows:

[0033] 1. The cementing quality evaluation system for deep shale gas wells of this invention collects and processes construction parameters and logging data of shale gas wells through a data collection component, and transmits the processed data to a data set module. A strength assessment component acquires the logging data from the data set module and determines the first interface cementing strength and the second interface cementing strength based on the acquired data. The determined strength data is then sent to the data set module for later use. A network training component acquires the construction parameters and strength data from the data set module and trains a neural network based on the acquired data. Subsequently, a structure determination component determines the neural network structure, and a real-time evaluation component is used to evaluate the cementing quality based on the determined neural network structure. This effectively solves the problems of long evaluation time, high cost, and inaccurate evaluation results in existing shale gas well cementing quality evaluation methods.

[0034] 2. The cementing quality evaluation system and method of this invention for deep shale gas wells can promptly detect and resolve cementing quality problems, improve the evaluation efficiency and accuracy of cementing quality, and also conduct comprehensive analysis and decision-making on the evaluation results of neural networks, which helps to more comprehensively evaluate cementing quality and take more reasonable measures to improve cementing quality.

[0035] 3. The real-time evaluation component of the present invention includes a threshold alarm setting module connected to the real-time quality evaluation module. When the result output by the neural network exceeds or falls below the threshold set in the threshold alarm setting module, the system automatically triggers an alarm mechanism to remind the operator to pay attention and take measures, thereby enabling the operator to detect cementing quality problems in a timely manner.

[0036] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention will be apparent from the effects described in the description and the accompanying drawings. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the cementing quality evaluation system for deep shale gas wells according to the present invention;

[0038] Figure 2 This is a flowchart of the data collection component in this invention;

[0039] Figure 3 This is a flowchart of the strength assessment component in this invention;

[0040] Figure 4 This is a flowchart of the network training component in this invention;

[0041] Figure 5 This is a flowchart of the structural determination component in this invention;

[0042] Figure 6 This is a flowchart of the real-time evaluation component in this invention;

[0043] Figure 7 This is a flowchart of the cementing quality evaluation method for deep shale gas wells according to the present invention.

[0044] The components include: 1-Data Collection Module, 2-Intensity Assessment Module, 3-Network Training Module, 4-Structure Determination Module, 5-Real-time Evaluation Module, 6-Data Set Module, 11-Construction Parameter Collection Module, 12-Well Logging Data Collection Module, 13-Well Logging Data Processing Module, 14-Well Logging Data Fusion Module, 15-Well Logging Feature Output Module, 16-Data Center Processing Module, 17-Construction Parameter Output Module, 18-Well Logging Data Output Module, 21-Well Logging Data Extraction Module, 22-First Interface Evaluation Module, 23-Second Interface Evaluation Module, 24-Intensity Data Output Module, 31-Neural Network Selection Module, and 32-Data Information. The network consists of the following modules: 33 - Construction Parameter Extraction Module; 34 - Strength Data Extraction Module; 35 - Network Training Iteration Module; 36 - Network Structure Determination Module; 37 - Network Performance Evaluation Module; 38 - Network Saving and Loading Module; 41 - Network Training Analysis Module; 42 - Weight Threshold Adjustment Module; 43 - Performance Stability Check Module; 44 - Structural Principle Determination Module; 45 - Layer Number and Node Setting Module; 46 - Connection Relationship Determination Module; 51 - Real-time Network System Module; 52 - Neural Network Embedding Module; 53 - Real-time Quality Evaluation Module; 54 - Threshold Alarm Setting Module; 55 - Data Interface Display Module; and 56 - Result-Assisted Decision-Making Module. Detailed Implementation

[0045] 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, 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] This invention provides a cementing quality evaluation system for deep shale gas wells, such as... Figure 1 As shown, the system includes a data collection component 1 for collecting and processing construction parameters and well logging data. The data collection component 1 is connected to a data set module 6 for aggregating data. The data set module 6 is connected to a strength assessment component 2 for evaluating the strength of cement sheath bonding and a network training component 3 for training neural networks. The strength data processed by the strength assessment component 2 is sent back to the data set module 6 for later use. The network training component 3 is connected to a structure determination component 4 for determining the structure of the neural network. The structure determination component 4 is connected to a real-time evaluation component 5 for evaluating the cementing quality of shale gas wells. The network training component 3 is connected to the real-time evaluation component 5.

[0047] The working process between data collection component 1, data set module 6, strength assessment component 2, network training component 3, structure determination component 4, and real-time evaluation component 5 is as follows: Data collection component 1 collects and processes the construction parameters and logging data of the shale gas well, and transmits the processed data to data set module 6. Strength assessment component 2 obtains the construction parameters from data set module 6, and determines the first interface cementing strength and the second interface cementing strength based on the obtained data. The determined strength data is then transmitted to data set module 6 for later use. Network training component 3 obtains the construction parameters and strength data from data set module 6, and trains the neural network based on the obtained data. Subsequently, the structure determination component 4 determines the neural network structure, and the real-time evaluation component 5 evaluates the cementing quality based on the determined neural network structure.

[0048] Data collection component 1, such as Figure 2 As shown, the system includes a construction parameter collection module 11 for collecting construction parameters and a well logging data collection module 12 for collecting well logging data. Construction parameters include cement slurry mix ratio, pumping pressure, pumping time, well depth, formation characteristics, etc., while well logging data includes data measured using acoustic logging, electromagnetic logging, density logging, and neutron logging technologies. The well logging data collection module 12 is connected to a well logging data processing module 13 for preprocessing the well logging data, including noise reduction, filtering, and calibration. The well logging data processing module 13 is connected to a well logging data fusion module 14 for fusion of well logging data. 4. The data is fused using methods such as Kalman filtering or weighted averaging. The logging data fusion module 14 is connected to a logging feature output module 15 for extracting characteristic parameters related to the cement sheath bonding strength from the fused data. The logging feature output module 15 is connected to a data center processing module 16 for data matching. The data center processing module 16 is connected to the construction parameter collection module 11. The data center processing module 16 is also connected to a construction parameter output module 17 for outputting construction parameters and a logging data output module 18 for outputting logging data. The construction parameter output module 17 and the logging data output module 18 are connected to the data set module 6.

[0049] The working process of data collection component 1 is as follows: construction parameter collection module 11 collects construction parameters, well logging data collection module 12 collects well logging data, and transmits the collected well logging data to well logging data processing module 13 for preprocessing. The preprocessed well logging data is transmitted to well logging data fusion module 14 for data fusion. The fused well logging data is transmitted to well logging feature output module 15 for feature extraction. The extracted feature parameters are transmitted to data center processing module 16. Data center processing module 16 performs matching processing on the construction parameters and the processed well logging data. The matched construction parameters and well logging data are transmitted to construction parameter output module 17 and well logging data output module 18, respectively, and then transmitted to data collection module 6 through construction parameter output module 17 and well logging data output module 18.

[0050] Strength assessment component 2, such as Figure 3 As shown, it includes a logging data extraction module 21 connected to the data set module 6. The logging data extraction module 21 is used to acquire logging data from the data set module 6. The logging data extraction module 21 is connected to a first interface evaluation module 22 for evaluating the bonding strength of the first interface between the cement sheath and the casing, and a second interface evaluation module 23 for evaluating the bonding strength of the second interface between the cement sheath and the bottom layer. The first interface evaluation module 22 is connected to a strength data output module 24, and the strength data output module 24 is connected to the second interface evaluation module 23.

[0051] The working process of the strength assessment component 2 is as follows: the well logging data extraction module 21 acquires the well logging data in the data set module 6, and transmits the acquired well logging data to the first interface assessment module 22 and the second interface assessment module 23 to determine the first interface bonding strength and the second interface bonding strength. The determined strength data is then transmitted to the strength data output module 24, and the strength data output module 24 transmits the determined strength data to the data set module for later use.

[0052] Network training component 3, such as Figure 4As shown, the system includes a construction parameter extraction module 33 for extracting construction parameters from the data set module 6 and an intensity data extraction module 34 for extracting intensity data from the data set module 6. The intensity data extraction module 34 is connected to a data information processing module 32, which performs data cleaning, normalization, and standardization operations on the extracted data. The data information processing module 32 is connected to a neural network selection module 31 and a network training iteration module 35. The neural network selection module 31 selects a suitable network, such as a BP neural network or a convolutional neural network, based on the data characteristics in the data set module 6. The network training iteration module 35 is connected to a network structure determination module 36. The network training component 3 includes a network performance evaluation module 37 connected to the network structure determination module 36, and the network performance evaluation module 37 is connected to a network storage and loading module 38 for storing the neural network.

[0053] The working process of network training component 3 is as follows: construction parameter extraction module 33 extracts construction parameters from data set module 6, strength data extraction module 34 extracts strength data from data set module 6, neural network selection module 31 selects a suitable network, the extracted construction parameters and strength data are transmitted to data information processing module 32 for processing, and the processed data is transmitted to network training iteration module 35 to train the neural network selected by neural network selection module 31. During the training process, through forward propagation, the input data is processed by the neural network to obtain the output result, and the output result is transmitted to network performance evaluation module 37 for evaluation.

[0054] Structure determines component 4, such as Figure 5 As shown, it includes a network training and analysis module 41, which is connected to the network structure determination module 36. The network training and analysis module 41 is connected to a structure determination principle module, a layer and node setting module 45, and a connection relationship determination module 46. The network training and analysis module 41 is also connected to a weight threshold adjustment module 42 for optimizing the performance of the neural network and a performance stability detection module for detecting the performance of the neural network.

[0055] The working process of the structure determination component 4 is as follows: The network training analysis module 41 obtains the neural network training status transmitted by the network structure determination module 36, and adjusts the weights and thresholds of the network through the weight and threshold adjustment module 42 using the backpropagation algorithm based on the training status to minimize the loss function and optimize network performance. The performance stability check module 43 analyzes the network performance changes, including indicators such as accuracy and loss function, to ensure that the network is stable and no longer has significant performance fluctuations. The structure principle determination module 44 and the layer and node setting module 45 ensure that the number of nodes in the input layer matches the dimension of cementing construction parameters and logging data. The number of layers and nodes in the hidden layer is adjusted according to the complexity of the training data and the performance of the network, and the number of nodes in the output layer matches the dimension of the cementing quality evaluation results. The connection relationship determination module 46 determines the connection relationship between nodes in each layer, including fully connected and convolutional connections, to construct a complete neural network model. The complete neural network model is then transmitted to the network saving and loading module 38 for saving.

[0056] Real-time evaluation component 5, such as Figure 6 As shown, the system includes a real-time network system module 51 capable of receiving and processing well logging data in real time. The real-time network system module 51 is connected to a neural network embedding module 52 for embedding neural networks stored in the network storage and loading module 38. The neural network embedding module 52 is connected to a real-time quality evaluation module 53 for evaluating the cementing quality of deep shale gas wells. The real-time evaluation component 5 also includes a threshold alarm setting module 54 connected to the real-time quality evaluation module 53. The threshold alarm setting module 54 sets alarm thresholds. When the output result of the neural network exceeds or falls below these thresholds, the system automatically triggers an alarm mechanism to remind operators to pay attention and take measures. The real-time quality evaluation module 53 is also connected to a data interface display module 55 for displaying results. The data interface display module 55 visualizes the output results of the neural network in the form of charts, curves, etc. The data interface display module 55 is connected to a result-assisted decision-making module 56. The result-assisted decision-making module 56 compares the output results based on pre-stored expert knowledge and experience, and synchronously displays the processing opinions for similar results or problems.

[0057] The working process of the real-time evaluation component 5 is as follows: the neural network embedding module 52 loads the neural network embedded system stored in the network storage loading module 38. At the same time, a system capable of receiving and processing logging data in real time is set in the real-time network system module 51, and the neural network is embedded into the system through the neural network embedding module 52. Then, the cementing quality evaluation of the deep shale gas well is completed through the real-time quality evaluation module 53.

[0058] The working principle of a cementing quality evaluation system for deep shale gas wells is as follows: During use, the construction parameter collection module 11 collects data including cement slurry mix ratio, pumping pressure, pumping time, well depth, and formation characteristics; the logging data collection module 12 collects logging data measured by acoustic logging, electromagnetic logging, density logging, and neutron logging technologies; the logging data processing module 13 preprocesses all logging data, including noise reduction, filtering, and calibration; and the preprocessed data is then fused using the logging data fusion module 14, employing methods such as Kalman filtering or weighted averaging. Then, the well logging feature output module 15 extracts feature parameters related to the cement sheath bonding strength from the fused data and transmits them to the data center processing module 16. The data center processing module 16 further matches and processes the construction parameters and well logging data, and transmits the construction parameters and well logging data to the data set module 6 for later use through the construction parameter output module 17 and the well logging data output module 18, respectively. By fusing and analyzing data obtained from multiple well logging techniques, more comprehensive and accurate cement sheath bonding information is obtained, improving the accuracy of the subsequent neural network output results. By saving the network... The neural network stored in the loading module 38 is loaded into the neural network embedding module 52. Simultaneously, a system capable of receiving and processing well logging data in real time is set up in the real-time network system module 51, and the neural network is embedded into this system through the neural network embedding module 52. Then, the real-time quality evaluation module 53 receives the well logging data in real time and transmits it to the neural network for processing, analysis, and output. This provides a low-cost and time-efficient evaluation technology for shale gas well cementing quality. Furthermore, by setting alarm thresholds in the threshold alarm setting module 54, the system automatically triggers an alarm mechanism when the neural network output exceeds or falls below these thresholds, alerting operators to take action. Simultaneously, the data interface display module 55 visualizes the neural network output in the form of charts, curves, etc. Based on the results, the result-assisted decision-making module 56 compares the results with pre-stored expert knowledge and experience, simultaneously displaying processing opinions for similar results or problems. This enables timely detection and resolution of cementing quality issues, improving evaluation efficiency and accuracy. Furthermore, comprehensive analysis and decision-making based on the neural network's evaluation results help to more comprehensively assess cementing quality and take more reasonable measures to improve it.

[0059] This invention also discloses a cementing quality evaluation method for deep shale gas wells, implemented based on the aforementioned cementing quality evaluation system for deep shale gas wells, such as... Figure 7 As shown, it includes the following steps:

[0060] Step (1): Obtain the construction parameters and logging data of the shale gas well, and process the construction parameters and logging data; specifically, this includes:

[0061] Step (101): Obtain the construction parameters and logging data of the shale gas well; specifically, the construction parameters and logging data are collected by the construction parameter collection module 11 and the logging data collection module 12 respectively. The construction parameters include the cement slurry ratio, pumping pressure, pumping time, well depth, formation characteristics, etc., and the logging data includes data measured by sonic logging technology, electromagnetic logging technology, density logging technology and neutron logging technology, etc.

[0062] Step (102): Preprocess, fuse, and extract feature parameters from the acquired logging data; specifically, the collected logging data is preprocessed by the logging data processing module 13, the preprocessed logging data is transmitted to the logging data fusion module 14 for data fusion, and the fused logging data is transmitted to the logging feature output module 15 for feature extraction; wherein, the preprocessing includes noise reduction, filtering, calibration, etc.

[0063] Step (103): Match the construction parameters with the well logging data after feature extraction; specifically, the extracted feature parameters are transmitted to the data center processing module 16, the data center processing module 16 matches the construction parameters and the processed well logging data, and transmits the construction parameters and well logging data to the data set module 6 for later use through the construction parameter output module 17 and the well logging data output module 18 respectively.

[0064] Step (2): Determine the bonding strength of the first interface and the bonding strength of the second interface based on the logging data; specifically, evaluate the bonding strength of the cement sheath, extract logging data from the data set module 6 through the logging data extraction module 21, evaluate the bonding strength of the first interface between the cement sheath and the casing through the first interface evaluation module 22, evaluate the bonding strength of the second interface between the cement sheath and the bottom layer through the second interface evaluation module 23, and transmit the bonding strength evaluation data of the first interface and the second interface to the data set module 6 through the strength data output module 24;

[0065] Step (3): Train the neural network based on construction parameters, the bonding strength of the first interface, and the bonding strength of the second interface; specifically, this includes:

[0066] Step (301): Select a neural network based on the data features of construction parameters and well logging data; specifically, the neural network selection module 31 selects a suitable network such as a BP neural network or a convolutional neural network based on the data features in the data set module 6.

[0067] Step (302): The selected neural network is trained based on construction parameters, well logging data, first interface bonding strength and second interface bonding strength; specifically, construction parameters and strength data are extracted from the data set module 6 by the construction parameter extraction module 33 and the strength data extraction module 34 respectively, and transmitted to the data information processing module 32 for data cleaning, normalization and standardization. Then, the neural network selected by the neural network selection module 31 is trained in the network training iteration module 35 by the extracted and processed data. During the training process, the input data is passed through the neural network to obtain the output result through forward propagation, and the output result is transmitted to the network performance evaluation module 37 for evaluation and comparison with the real result to see if it meets the set conditions.

[0068] Step (4): Adjust the training parameters in the neural network training and determine the neural network structure; specifically, the network structure determination module 36 synchronously transmits the training process data to the network training analysis module 41. The network training analysis module 41 adjusts the weights and thresholds of the network through the weight threshold adjustment module 42 using the backpropagation algorithm according to the training situation to minimize the loss function and optimize the network performance. The performance stability check module 43 analyzes the network performance changes, including indicators such as accuracy and loss function, to ensure that the network is stable and no longer has obvious performance fluctuations. The structure principle determination module 44 and the layer node setting module 45 ensure that the number of nodes in the input layer matches the dimension of cementing construction parameters and logging data. The number of layers and nodes in the hidden layer are adjusted according to the complexity of the training data and the performance of the network, and the number of nodes in the output layer matches the dimension of the cementing quality evaluation results. The connection relationship determination module 46 determines the connection relationship between the nodes of each layer, including fully connected, convolutional connection, etc., to build a complete neural network model. Finally, the trained neural network is saved through the network saving and loading module 38.

[0069] Step (5): Cementing quality evaluation is performed based on the determined neural network structure. Specifically, the neural network stored in the network storage and loading module 38 is loaded into the neural network embedding module 52. At the same time, a system capable of receiving and processing logging data in real time is set in the real-time network system module 51, and the neural network is embedded into the system through the neural network embedding module 52. Then, the logging data is received in real time through the real-time quality evaluation module 53 and transmitted to the neural network for processing, analysis and output. An alarm threshold is set in the threshold alarm setting module 54. When the output result of the neural network exceeds or falls below these thresholds, the system automatically triggers the alarm mechanism to remind the operator to pay attention and take measures. At the same time, the output result of the neural network is visualized in the form of charts, curves and other forms through the data interface display module 55. Based on the result, the result auxiliary decision-making module 56 compares it with the pre-stored expert knowledge and experience, and displays the processing opinions of similar results or problems synchronously.

[0070] Although specific embodiments of the present invention have been described in detail with reference to the accompanying drawings, this should not be construed as limiting the scope of protection of this patent. Various modifications and variations that can be made by those skilled in the art without inventive effort within the scope described in the claims still fall within the scope of protection of this patent.

Claims

1. A cement job quality evaluation system for a deep shale gas well, characterized by: The system includes a data collection component connected to a data set module for collecting data. The data set module is connected to a strength assessment component for evaluating the strength of cement sheath bonding and a network training component for training a neural network. The network training component is connected to a structure determination component for determining the structure of the neural network. The structure determination component is connected to a real-time evaluation component for evaluating the cementing quality of shale gas wells. The network training component is connected to the real-time evaluation component.

2. The cement job quality evaluation system for a deep shale gas well of claim 1, wherein: The data collection component includes a construction parameter collection module and a well logging data collection module. The well logging data collection module is connected to a well logging data processing module for processing well logging data. The well logging data processing module is connected to a well logging data fusion module for well logging data fusion. The well logging data fusion module is connected to a well logging feature output module for extracting feature parameters. The well logging feature output module is connected to a data center processing module. The data center processing module is connected to the construction parameter collection module. The data center processing module is also connected to a construction parameter output module for outputting construction parameters and a well logging data output module for outputting well logging data. The construction parameter output module and the well logging data output module are connected to a data set module.

3. The cement job quality evaluation system for a deep shale gas well of claim 2, wherein: The construction parameters collected by the construction parameter collection module include cement slurry mix ratio, pumping pressure, pumping time, well depth, and formation characteristics.

4. The cement job quality evaluation system for a deep shale gas well of claim 1, wherein: The strength assessment component includes a well logging data extraction module connected to a data set module. The well logging data extraction module is connected to a first interface assessment module and a second interface assessment module. The first interface assessment module is connected to a strength data output module, and the strength data output module is connected to the second interface assessment module.

5. The cement job quality evaluation system for a deep shale gas well of claim 1, wherein: The network training component includes a construction parameter extraction module and a strength data extraction module. The strength data extraction module is connected to a data information processing module. The data information processing module is connected to a neural network selection module and a network training iteration module. The network training iteration module is connected to a network structure determination module for determining the neural network.

6. The cement job quality evaluation system for a deep shale gas well of claim 5, wherein: The network training component includes a network performance evaluation module connected to the network structure determination module, and the network performance evaluation module is connected to a network saving and loading module for storing the neural network.

7. The cement job quality evaluation system for a deep shale gas well of claim 1, wherein: The structure determination component includes a network training and analysis module, which is connected to a structure determination principle module, a layer and node setting module, and a connection relationship determination module.

8. The cementing quality evaluation system for deep shale gas wells as described in claim 7, characterized in that: The network training and analysis module is also connected to a weight threshold adjustment module for optimizing neural network performance and a performance stability detection module for detecting neural network performance.

9. A cementing quality evaluation system for deep shale gas wells as described in claim 1, characterized in that: The real-time evaluation component includes a real-time network system module, which is connected to a neural network embedding module, and the neural network embedding module is connected to a real-time quality evaluation module for cementing quality evaluation of deep shale gas wells.

10. The cement job quality evaluation system for a deep shale gas well of claim 9, wherein: The real-time evaluation component includes a threshold alarm setting module connected to the real-time quality evaluation module.

11. The cement job quality evaluation system for a deep shale gas well of claim 9, wherein: The real-time quality evaluation module is connected to a data interface display module for displaying results, and the data interface display module is connected to a result-assisted decision-making module.

12. A method for evaluating cementing quality of a deep shale gas well, implemented based on the cementing quality evaluation system for the deep shale gas well according to any one of claims 1-11, characterized in that, Includes the following steps: Acquire construction parameters and logging data for shale gas wells, and perform data processing on the construction parameters and logging data; Based on well logging data, determine the bonding strength of the first interface and the bonding strength of the second interface; Neural network training is performed based on construction parameters, the bonding strength of the first interface, and the bonding strength of the second interface. Adjusting training parameters during neural network training to determine the neural network structure; Cementing quality is evaluated based on the determined neural network structure.

13. The method for evaluating cementing quality of a deep shale gas well according to claim 12, characterized in that, The acquisition of shale gas well construction parameters and logging data, and the processing of these parameters and data, include: Obtain construction parameters and logging data for shale gas wells; The acquired well logging data is preprocessed, fused, and its characteristic parameters are extracted. The construction parameters are matched with the well logging data after feature extraction.

14. The method for evaluating cementing quality of a deep shale gas well according to claim 12, characterized in that, The neural network training based on construction parameters, first interface bonding strength, and second interface bonding strength includes: Neural network selection based on data features of construction parameters; The selected neural network is trained based on construction parameters, the bonding strength of the first interface, and the bonding strength of the second interface.

15. The method for evaluating cementing quality of a deep shale gas well according to claim 14, characterized in that, The performance of the selected neural network is evaluated during the training process.

16. The method for evaluating cementing quality of a deep shale gas well of claim 12, wherein, The regulation of training parameters in the training of the neural network to determine the neural network structure includes: regulating the weights, thresholds, input layer, output layer and hidden layer of the neural network to determine the neural network structure.

Images