Hydraulic fracturing operation anomaly and fracturing effect coordination identification method and system

Abstract

This invention discloses a method and system for collaborative identification of hydraulic fracturing construction anomalies and fracturing effectiveness. It collects geological data of the fracturing reservoir and multi-source construction data, and performs standardized preprocessing. The fracturing construction process is divided into multiple core stages, and real-time stage status identification is performed. Multi-source dimensional features are extracted for each identification stage, and fracturing initiation confidence index, expansion effectiveness index, and anomaly risk index are constructed and calculated. Based on the three core indices and stage status variables, the types of construction anomalies are classified and identified. Simultaneously, a comprehensive score for fracturing effectiveness is calculated and effectiveness levels are assigned. Finally, matching construction and treatment suggestions are output in a linked manner. This invention achieves collaborative identification of rapid anomaly conditions and real-time evaluation of fracturing effectiveness during fracturing construction. It relies only on low-cost monitoring parameters easily obtained in underground coal mines. Through phased dynamic modeling, it significantly reduces the cross-stage misjudgment rate, fully meeting the on-site engineering needs of "construction, identification, and adjustment simultaneously" in underground coal mine fracturing.

Description

Technical Field

[0001] This invention belongs to the fields of hydraulic fracturing, unconventional reservoir permeability enhancement, engineering monitoring and intelligent diagnosis technology. Specifically, it relates to a method and system for rapid identification of abnormal working conditions, real-time evaluation of fracturing effectiveness and joint response suggestions in the hydraulic fracturing construction process. Background Technology

[0002] Hydraulic fracturing in coal mines is a core engineering measure to improve coal seam permeability, enhance gas extraction, reduce the risk of coal and gas outbursts, and achieve pressure relief and permeability enhancement. Compared with surface shale oil and gas fracturing, underground hydraulic fracturing in coal mines has distinct engineering characteristics: First, the construction space is limited, and equipment configuration is simplified. The underground site is small, making it difficult to deploy complex monitoring systems. Often, only limited parameters such as orifice pressure, discharge rate, cumulative injection volume, and sealing section pressure can be stably obtained. Second, the geological conditions are complex and highly heterogeneous. Coal body fractures are well-developed, with obvious interlayering of soft and hard layers, and uneven distribution of local structures and primary fractures, resulting in large dispersion in fracturing response. The same construction parameters show significant differences in performance in different borehole sections. Third, the construction process is highly dynamic. Fracturing is not a single static process but a dynamic process that includes pressurization, fracturing initiation, propagation, pump shutdown, closure, and inter-section transitions. The physical mechanisms and anomaly manifestations differ at different stages, requiring on-site construction to focus more on real-time status identification and dynamic decision-making.

[0003] When hydraulic fracturing in coal mines is used for gas drainage and permeability enhancement, or for directional long borehole fracturing of coal seams or roof and floor, the requirements for construction safety, effectiveness, and real-time decision-making are extremely high. However, on-site monitoring conditions are limited, necessitating real-time identification of typical operating conditions during construction, such as water leakage, failure to fracture, cross-cutting, roadway leakage, borehole failure, abnormal pressure drop, insufficient fracturing, or obstructed propagation. Engineering practice shows that the most critical requirement for underground fracturing is not post-construction explanation, but real-time judgment during construction: whether fracturing has initiated, whether the fracture is effectively propagating, whether leakage / cross-cutting / borehole failure has occurred, whether the current section is worth continuing fracturing, and what adjustment measures should be taken.

[0004] In the existing technology, research on fracturing monitoring and diagnosis is mostly focused on the field of surface oil and gas fracturing. The main technical approaches include: interpreting the initiation pressure, fracture pressure and closure pressure based on the characteristics of the pressure curve; performing pressure drop analysis based on the pressure decay after pump shutdown; identifying fracture propagation based on high-cost monitoring methods such as microseismic, tiltmeter, and acoustic emission; and conducting post-operation review based on the results of a single operation.

[0005] However, existing technologies have the following significant shortcomings: (1) Emphasis on post-construction interpretation, neglect of real-time on-site identification. Most methods are only applicable to the interpretation of fracturing curves or the analysis of post-construction effects after construction is completed, and cannot meet the core on-site requirements of "construction, identification and adjustment at the same time" during underground fracturing construction in coal mines.

[0006] (2) Emphasis on single parameters and neglect of multi-source fusion. Many methods rely too much on a single pressure curve or fixed empirical threshold, which makes it difficult to adapt to the complex and unsteady working conditions in coal mines. They are not good at distinguishing hidden anomalies such as cross-sections, near-well leakage, and sealing failure, resulting in a high misjudgment rate.

[0007] (3) Emphasis on static judgment and neglect of stage evolution analysis. Existing methods often fail to clearly divide and differentiate the stages of fracturing construction, which can easily lead to misjudging normal responses at different stages as abnormal ones, or misjudging abnormal conditions as normal fluctuations; (4) High cost monitoring, but poor underground deployability. Monitoring technologies commonly used in surface fracturing, such as microseismic and tiltmeter monitoring, are costly to implement in coal mines, have poor environmental adaptability, and are difficult to deploy continuously, making it difficult to form a low-cost real-time identification system adapted to underground scenarios.

[0008] (5) Lack of an integrated solution for anomaly identification, effectiveness evaluation, and handling recommendations. On-site engineering requires not only anomaly identification results, but also a clear understanding of the impact of anomalies on fracturing effectiveness, the effectiveness of the current fracturing section, and corresponding adjustment measures. Existing technologies lack a systematic design for this.

[0009] In summary, existing technologies cannot adapt to the engineering characteristics and on-site requirements of hydraulic fracturing in coal mines. There is an urgent need to propose a technical solution that is oriented towards the on-site construction process of segmented fracturing in coal mines, aims to solve practical engineering problems, takes into account both mechanism explanation and artificial intelligence recognition capabilities, and can be deployed in real time under low monitoring configuration conditions to identify construction anomalies early and simultaneously judge fracturing effectiveness. Summary of the Invention

[0010] To address the problems existing in the prior art, this invention provides a method and system for collaboratively identifying anomalies and fracturing effectiveness in coal mine hydraulic fracturing operations. It achieves real-time perception of the fracturing operation status by relying solely on readily available downhole parameters such as orifice pressure, discharge rate, cumulative injection volume, pump shutdown pressure response, and sealing section pressure. It avoids cross-stage misjudgments through phased dynamic modeling; it rapidly identifies typical abnormal conditions in real time; it simultaneously evaluates the effectiveness of fracturing initiation, propagation, pressure maintenance, and overall fracturing effectiveness of the fracturing section; it outputs coordinated response suggestions that can directly serve on-site operations; and it introduces a lightweight artificial intelligence model to enhance the generalization and adaptability of solutions in complex downhole scenarios.

[0011] To achieve the above objectives, the technical solution adopted by this invention is: a method for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness, comprising the following steps: S1. Data Acquisition and Standardized Preprocessing: Collect geological data of the target fracturing reservoir, and simultaneously collect multi-source real-time data during the fracturing operation. Preprocess the collected multi-source real-time data to obtain a standardized input vector.

[0012] S2. Stage-by-stage state identification: The entire fracturing operation process is divided into at least five core stages. Based on the pre-processed standardized data, the current core stage of the operation is identified in real time, and the stage state variable S(t) is generated. The five core stages include the injection and pressurization stage, the fracturing initiation stage, the fracture propagation stage, the pump stop pressure holding / closing response stage, and the inter-stage transition stage.

[0013] S3. Multi-source feature extraction in stages: For the current core construction stage identified in step S2, multi-source features corresponding to the stage are extracted from the standardized input vector. The multi-source features include at least the basic parameter features, pressure displacement coupling features, pump stoppage attenuation features, internal and external pressure difference coefficients, and crack propagation response features.

[0014] S4. Collaborative Construction of Core Indices: Based on the multi-source features extracted in step S3, three types of core quantitative indices are constructed and calculated in real time, namely, the crack initiation confidence index I. f Extended Effective Index I e Abnormal Risk Index I a .

[0015] S5. Classification of Construction Anomalies: Based on the three core quantitative indices obtained in step S4 and the current stage state variable S(t) obtained in step S2, the anomaly type classification of the current fracturing construction is performed, and the corresponding anomaly type identification results are output.

[0016] S6. Comprehensive Evaluation of Fracturing Effectiveness: Based on the three core quantitative indices obtained in step S4, combined with the reasonable closure / pressure holding index I. c Calculate the comprehensive score of fracturing effectiveness for the current fracturing section, and classify the fracturing effectiveness level based on the score results.

[0017] S7. Action Recommendation Linkage Output: Based on the anomaly type identification results output in step S5 and the fracturing effectiveness level obtained in step S6, linkage output of construction action recommendations matching the current working conditions.

[0018] Furthermore, in step S1, the geological data of the target fracturing reservoir includes at least lithology, hydrology, structure, fracture zone, and original geostress distribution; the multi-source real-time data during the fracturing operation includes at least: real-time orifice pressure data P(t), real-time fracturing operation discharge data Q(t), cumulative injection volume V(t), real-time sand ratio sequence S(t), and real-time pressure data P in the fracturing section. s (t), Pressure decay sequence P after pump shutdown sh(t), return liquid volume / return water volume sequence Q b (t), water ingress signal W(t) at the wellhead; additionally, if a microseismic monitoring system is present at the fracturing site, the microseismic monitoring signal A(t) is collected as an auxiliary signal to further increase the amount of data acquired. The real-time sand ratio sequence S... v The formula for (t) is: in, Real-time volumetric flow rate of the proppant; This represents the real-time volumetric flow rate of the fracturing fluid.

[0019] The preprocessing includes at least time synchronization, outlier removal, smoothing filtering, unified sampling interval reconstruction, condition slicing, and stage labeling; the expression for the standardized input vector is: The normalization process for each variable in the standardized input vector is expressed as follows: in, The historical mean of the corresponding variable. The historical standard deviation of the corresponding variable. To prevent small quantities with a denominator of zero.

[0020] Furthermore, in step S2, the expression for the stage state variable S(t) is: Among them, S1 corresponds to the injection and pressurization stage, S2 corresponds to the crack initiation stage, S3 corresponds to the crack propagation stage, S4 corresponds to the pump stop pressure holding / closing response stage, and S5 corresponds to the inter-segment transition stage.

[0021] The real-time stage identification specifically includes the following steps: S21. Rule Prior Identification: The initial stage is determined based on prior rules of displacement change, pressure derivative, pump stoppage event, and injection volume increase. The specific determination rules are as follows: If real-time displacement And real-time pressure change rate If so, it is determined that the injection and pressurization phase has begun, in which... This is the injection threshold.

[0022] If a local pressure peak occurs and is accompanied by a sudden drop in the pressure slope, it is determined that the crack initiation candidate stage has begun.

[0023] If pumping continues and the pressure enters the fluctuation plateau range, it is determined that the crack has entered the crack propagation stage.

[0024] If real-time displacement If the pressure decreases rapidly, it is determined that the pump has entered the pressure holding / closing response phase.

[0025] If there is packer movement, tool string switching, or a continuous lack of fluid injection, it is determined that the inter-segment switching phase has begun.

[0026] S22. Time series model correction: The initial stage identification results are smoothly corrected using existing methods (such as lightweight hidden Markov model HMM or gated cyclic unit GRU). Unreasonable stage jumps are constrained by the state transition matrix (such as prohibiting direct jumps from the injection pressurization stage to the pump stop pressure holding / closing response stage), and the final stage state variable S(t) is output.

[0027] Furthermore, in step S3, the specific calculation method for the multi-source features is as follows: (1) Basic parameter characteristics: including real-time pressure change rate Real-time displacement change rate Cumulative injection volume change rate Pressure derivative The specific formula is as follows: ; ; ; ; (2) Pressure-displacement coupling characteristics include pressure increase coefficient per unit displacement. Unit volume pressure increase coefficient Pressure fluctuation index The specific formula is as follows: ; ; ; (3) Pump shutdown attenuation characteristics: Fitting formula for pressure attenuation after pump shutdown ,in The instantaneous pressure at which the pump stops. To fit the closed characteristic pressure, This is the attenuation coefficient.

[0028] (4) Internal and external pressure difference coefficient: sealing pressure difference When this value continues to decrease abnormally and is accompanied by water coming into the orifice and an increase in liquid return, it indicates an increased probability of orifice failure, cross-hole leakage, and tunnel leakage.

[0029] (5) Crack propagation response characteristics: including propagation plateau stability ,in To expand the platform's pressure standard deviation during the expansion phase, To extend the platform pressure range; injection absorption index At similar displacements, the higher this value and the more stable the pressure plateau, the more fully the fracture absorbs and propagates liquid.

[0030] Furthermore, in step S4, the crack initiation confidence index I f The expression used to characterize the actual fracturing state of the current fracturing segment is: in, The characteristic of continuous pressure increase before the peak point The magnitude of the sudden drop in pressure after the peak point, The significance of the pressure derivative changing from positive to negative. For the sudden change response of unit injection volume, For auxiliary signal triggering characteristics (microseismic / orifice response); the weighting coefficients satisfy... Each component is normalized and its value range is [0,1].

[0031] when At that time, it was determined that crack initiation had been successful; when When the cracking is deemed insufficient or the initiation is uncertain, the following conditions are met: The threshold for determining crack initiation can be determined based on historical data from similar work areas.

[0032] The extended efficiency index I e Used to characterize the effective fracture propagation state and distinguish between effective fracture propagation and near-wellbore pressure buildup, its expression is: in, To extend the stability of the pressure platform during the extended phase, Pressure maintenance capacity per unit injection volume To ensure the rationality of the closed-loop response after pump shutdown For the hysteresis characteristics of return-injection coupling, The response characteristics of adjacent holes or auxiliary signals; the weighting coefficients satisfy... ;I e A higher value indicates sufficient crack propagation, reasonable liquid absorption, and normal pump shutdown response; I e The lower the value, the more likely it is to indicate hindered propagation, near-wellbore compaction, fluid stagnation in the near-wellbore zone, and insufficient fracturing effectiveness; when... At that time, it was determined that the crack had effectively expanded, among which The threshold for determining the effective propagation of cracks.

[0033] The abnormal risk index I a The expression used to characterize the abnormal risk level of the current fracturing operation is as follows: in, Enhanced characteristics for water seepage or abnormal backflow at the orifice. The characteristic of abnormally reduced sealing pressure difference The characteristic of abnormally rapid pressure drop Characteristics of pressure holding failure For abnormal fluctuation / pulsation enhancement characteristics, For auxiliary equipment load anomaly characteristics; the weighting coefficients satisfy ;I a The value range is [0,1], with higher values ​​indicating a greater risk of anomalies. When an anomaly warning is triggered, among which... The threshold for determining abnormal risks can be calibrated according to on-site safety requirements.

[0034] Furthermore, in step S5, the abnormal types of fracturing conditions include at least: normal pressure initiation, successful fracturing initiation, difficult fracturing / no fracturing, effective fracture propagation, insufficient fracture propagation / near-hole compaction, roadway leakage, and excessive filtration loss, such as perforation of fractured zones / structures, hole sealing failure / tool ​​string packer damage (poor setting, bulging of rubber sleeves, etc.), cross-flow / channeling, abnormal pressure drop, normal pump shutdown and closure, multiple fractures / fracture deflection (manifested as large pressure fluctuations, fracture extension direction deviating from the design, resulting in insufficient effective stimulation volume), perforation of aquifers / water bodies (characterized by backflow volume), abnormal gas, and abnormal mine pressure; among the abnormal conditions of difficult fracturing and insufficient fracture propagation, sand plugging is included, which is manifested as proppant accumulation in the sand-carrying fluid in the fracture, a sudden increase in construction pressure, pump pressure overload, and even inability to continue injection.

[0035] The anomaly type classification and identification adopts a fusion discrimination method of "mechanism rule screening + AI model subdivision", specifically: ① First, eliminate obviously normal and obviously abnormal operating conditions through mechanism rules.

[0036] ② A machine learning model is used to classify complex boundary conditions in detail and output the results of anomaly type identification.

[0037] The machine learning model includes any one of the following: decision tree, XGBoost, lightweight fully connected neural network, and 1D-CNN+GRU fusion network.

[0038] Furthermore, in step S6, the expression for the comprehensive score of fracturing effectiveness is: in, Let be the weight coefficient, and satisfy... ; Closure / Pressure Holding Reasonable Index I c The value is calculated based on the pressure decay characteristics and pressure holding time after pump shutdown, and the range is [0,1].

[0039] Based on the comprehensive scoring results, fracturing effectiveness is divided into five levels: A to E. ① Grade A: Score ≥ 0.85, indicating clear crack initiation, sufficient propagation, low risk of abnormality, and excellent construction effect.

[0040] ② Grade B: 0.70≤Score<0.85, crack initiation is clear, and the propagation is basically effective, with a slight risk.

[0041] ③ Grade C: 0.55≤Score<0.70, cracks have started, but the expansion is generally normal or locally abnormal, and the effect is moderate.

[0042] ④ Grade D: 0.40≤Score<0.55, construction abnormalities are obvious, and the current section has poor results.

[0043] ⑤ Grade E: Score < 0.40, no cracking or severe failure, requires immediate action.

[0044] Furthermore, in step S7, the construction treatment recommendations include at least: ① continuing pressure stabilization injection, ② reducing discharge rate to stabilize pressure, ③ increasing discharge rate to promote fracturing, ④ briefly stopping the pump and repressurizing, ⑤ checking the sealer and orifice seal, ⑥ identifying suspected cross-sections and suspending construction, ⑦ ending the current section and moving to the next section, ⑧ supplementing injection or repeating fracturing, and ⑨ immediately stopping construction if there is danger; a mapping relationship table between anomaly type, effectiveness level, and treatment recommendations is pre-established, and based on the anomaly type identification result in step S5 and the effectiveness level in step S6, the corresponding construction treatment recommendations are matched and output.

[0045] Furthermore, step S5 also includes specific identification of typical working conditions in coal mines: Firstly, the distinction between sealing failure and roadway leakage is as follows: Construct the discriminant function: in, For sealing pressure differential; This refers to the volume of returned liquid; For the pump shutdown pressure maintenance time, I f As the crack initiation confidence index, These are the weighting coefficients.

[0046] when When the sealing fails, it is determined to be a failure; when At that time, it was determined to be a leak in the tunnel, among which To distinguish and determine the threshold.

[0047] Secondly, segment / channel identification, specifically: in, For adjacent hole response intensity, To improve platform stability; , , For weighting coefficients; when When this occurs, a segment warning is triggered, in which... This is the threshold for segment determination.

[0048] On the other hand, the present invention provides a system for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness, used to implement the above-mentioned collaborative identification method, the system comprising: The data acquisition module is used to collect geological data of the target fracturing reservoir, as well as multi-source real-time data during the fracturing operation.

[0049] The preprocessing module communicates with the data acquisition module and is used to preprocess the acquired multi-source real-time data and output a standardized input vector.

[0050] The stage identification module communicates with the preprocessing module and is used to divide the entire fracturing construction process into stages. Based on the standardized input vector, it completes the real-time identification of the current construction stage and outputs the stage state variable S(t).

[0051] The feature extraction module communicates with the preprocessing module and the stage identification module respectively, and is used to extract multi-source features corresponding to the current construction stage from the standardized input vector.

[0052] The index calculation module, which communicates with the feature extraction module, is used to construct and calculate the crack initiation confidence index I in real time based on the extracted multi-source features. f Extended Effective Index I e Abnormal Risk Index I a Three core quantitative indices.

[0053] The anomaly classification module communicates with the stage identification module and the index calculation module respectively. It is used to classify and identify the anomaly types in fracturing construction based on three types of core quantitative indices and stage state variables S(t), and output the anomaly type identification results.

[0054] The performance evaluation module communicates with the index calculation module and is used to combine three types of core quantitative indices with the reasonable closure / holding pressure index I. c Calculate the comprehensive score of fracturing effectiveness and classify the effectiveness level.

[0055] The handling suggestion linkage module communicates with the anomaly classification module and the effectiveness evaluation module respectively, and is used to output matching construction handling suggestions based on the anomaly type identification results and fracturing effectiveness level.

[0056] The alarm display module is connected to the anomaly classification module, the effectiveness evaluation module, and the handling suggestion linkage module. It is used to display the construction status and identification results in real time, and to issue audible and visual alarm prompts when an anomaly warning is triggered.

[0057] The storage module is connected to each of the above modules and is used to store raw data, preprocessed data, feature data, model parameters, identification results and processing suggestions.

[0058] Compared with the prior art, the present invention has the following advantages: 1. This invention addresses the actual needs of hydraulic fracturing engineering in coal mines, enabling collaborative real-time identification of fracturing construction anomalies and evaluation of fracturing effectiveness. It does not merely provide post-fracturing explanations, but rather identifies anomalies, assesses effectiveness, and outputs handling suggestions in real time during construction. It fully adapts to the core on-site needs of "construction, identification, and adjustment simultaneously" in underground coal mine fracturing, and can directly serve frontline construction decision-making.

[0059] 2. This invention is adapted to the characteristics of underground coal mine construction and has the advantages of low cost and easy deployment. It mainly conducts analysis based on conventional monitoring parameters that are easily obtained underground, such as orifice pressure, discharge rate, injection rate, and sealing pressure. It does not rely heavily on high-cost monitoring methods such as microseismic and tiltmeter monitoring, thus solving the problems of poor underground adaptability and high deployment difficulty of existing technologies. It can be quickly promoted and applied in coal mines.

[0060] 3. This invention introduces a phased dynamic modeling mechanism, which divides the entire fracturing process into multiple core stages and performs differentiated analysis. This avoids the cross-stage misjudgment caused by the traditional method of "one-size-fits-all" analysis of the entire fracturing curve, significantly reduces the probability of confusion between normal and abnormal operating conditions, and improves the accuracy of the identification results.

[0061] 4. This invention is the first to construct three core quantitative indicators: crack initiation credibility index, expansion effectiveness index, and anomaly risk index. It realizes the simultaneous quantitative characterization of three core dimensions: "whether crack initiation occurs", "whether expansion is effective", and "whether anomaly exists". It forms a collaborative identification system for anomalies and effectiveness, filling the gap in the lack of an integrated quantitative identification system in the existing technology. It has strong engineering application value and feasibility.

[0062] 5. This invention adopts a fusion discrimination method of mechanistic rule screening + AI model subdivision, which not only ensures the mechanistic interpretability of the discrimination results and avoids the problem of low field acceptance of pure black box AI models, but also improves the fine classification and generalization capabilities under complex working conditions through artificial intelligence models. At the same time, the lightweight model can be deployed on downhole edge computing terminals to meet real-time requirements.

[0063] 6. This invention realizes the entire process of data acquisition, preprocessing, stage identification, feature extraction, index calculation, anomaly classification, effectiveness evaluation, and disposal suggestions. It can not only identify abnormal working conditions, but also simultaneously evaluate the impact of anomalies on fracturing effectiveness and output construction disposal suggestions that can be directly implemented. It forms a complete intelligent fracturing construction decision-making scheme, which greatly reduces the dependence on personnel experience in on-site construction and improves the safety, effectiveness, and standardization of hydraulic fracturing construction in coal mines. Attached Figure Description

[0064] Figure 1 This is a flowchart illustrating the overall process of the method in this invention.

[0065] Figure 2 This is a schematic diagram of a segmented fracturing tool string structure in an underground coal mine.

[0066] Figure 3 This is a schematic diagram of the system structure in this invention. Detailed Implementation

[0067] The present invention will be further described below.

[0068] Example 1: Real-time On-site Identification Method Based on Rules and Index Calculation like Figures 1 to 2 As shown, this embodiment is applied to directional long borehole segmented fracturing construction in a coal mine. It performs real-time identification of the target fracturing section during the construction process. The monitoring signals that can be collected on-site include borehole pressure P(t), displacement Q(t), cumulative injection volume V(t), real-time sand ratio sequence S(t), and sealing pressure P. s (t) and return liquid volume Q b (t), the specific implementation steps are as follows: Step 1: Data Acquisition and Standardized Preprocessing: Collect geological data of the target reservoir, including coal seam lithology, geostress distribution, and structural development; simultaneously collect real-time monitoring data during fracturing operations, use a 3-second moving average to smooth the pressure data, perform linear interpolation on missing sampling points, complete time synchronization and unified sampling interval reconstruction, obtain the standardized input vector X(t), and perform normalization processing.

[0069] Step 2, Phased State Identification: Based on the preprocessed data, phases are divided using rule-based prior identification: when Q(t) rises from 0 to above the set injection threshold and dP / dt>0, the injection pressurization phase is determined; when a pressure peak P is detected... peak And satisfy ,at the same time If the pressure is too high, the crack initiation stage is determined; if the pumping continues and the pressure enters the fluctuation plateau range, the crack propagation stage is determined.

[0070] Step 3: Multi-source feature extraction in stages: For the current core construction stage identified in Step 2, extract the multi-source features corresponding to the stage from the standardized input vector.

[0071] Step 4: Calculation of the crack initiation confidence index: In this embodiment, the weight configuration of the crack initiation confidence index is: w f1 =0.2、w f2 =0.3、w f3 =0.25、w f4 =0.15、w f5 =0.1, the calculation expression is: Based on calculations, the confidence index I for fracturing initiation in this fracturing section is... f =0.81, set the crack initiation judgment threshold θ f =0.65, therefore it is determined that the segment has successfully started cracking.

[0072] Calculation of the extended effective index: During the crack propagation stage, data from the pressure plateau interval are extracted, and the stability of the propagation plateau (e1=0.92), the pressure maintenance capability per unit injection volume (e2=0.85), and the rationality of the closure response after pump shutdown (e3=0.88) are calculated. The weights are configured as w. e1 =0.3、w e2 =0.3、w e3 =0.2、w e4 =0.1、w e5 =0.1, and the comprehensive calculation yields the extended effective index I. e =0.87, and the effective crack propagation judgment threshold θe=0.75 is set, indicating that the overall crack propagation in this segment is effective.

[0073] Step 5: Anomaly Risk Identification and Classification: In this embodiment, the increased water inflow signal at the orifice, the significant decrease in the sealing pressure differential, and the excessively rapid increase in backflow were detected simultaneously, and the anomaly risk index I was calculated. a =0.72, higher than the preset anomaly threshold θ a =0.60, triggering an abnormal warning; further calculation using the discriminant function of sealing failure and roadway leakage determined it to be "suspected roadway water leakage".

[0074] Step Six: Comprehensive Evaluation of Fracturing Effectiveness: Combining the Closure / Pressure Holding Rationality Index I c =0.82, with weights configured as ρ1=0.4, ρ2=0.3, ρ3=0.2, ρ4=0.1, calculate the overall score: Based on the scoring range, this segment is judged to be grade B, which is slightly better.

[0075] Step 7, Action Recommendation Output: Based on the anomaly type identification result (suspected tunnel water leakage) and effectiveness level (Level B), the system automatically outputs action recommendations: ① Appropriately reduce discharge rate and stabilize pressure; ② Continuously observe changes in sealing pressure; ③ If water seepage at the borehole continues to increase, suspend construction and inspect the sealing; ④ If the anomaly persists for more than 10 minutes, terminate construction in this section and proceed to the remedial procedure.

[0076] Example 2: Implementation of Anomaly Classification Based on Artificial Intelligence This embodiment constructs an artificial intelligence anomaly classification model based on historical data from multiple coal mine fracturing operation sections to achieve precise anomaly identification under complex working conditions. The specific implementation method is as follows: Sample set construction: Based on historical data from over 2000 fracturing operations across multiple coal mines, a labeled sample set D was constructed. Among them, X i For a multi-source parameter time series, S i For stage state variables, If i 、Ie i Ia i There are three types of core indices, y i The working conditions are manually labeled, including 15 categories such as successful crack initiation, no crack initiation, roadway leakage, borehole failure, and cross-section.

[0077] Model Input and Structure Design: A time window of length L=60 is selected, and each sample contains a multi-source parameter sequence of the most recent 60 time steps; a 1D-CNN+GRU fusion network model is adopted, with the following specific structure: 1D-CNN layer: Performs convolutional feature extraction on the input time series, outputting local temporal features. GRU layer: Models the temporal dependency of convolutional features and outputs global temporal features. Feature fusion layer: This layer combines temporal features with mechanistic features (If, Ie, Ia, decay coefficient, plateau stability, etc.). in, Mechanistic characteristics, including , , Attenuation coefficient, platform stability, etc.

[0078] Fully connected and classification layers: Outputting the probability of each class through the Softmax activation function: Model Training: The model is trained using a weighted cross-entropy loss function, with weights augmented for rare outlier types to address the imbalanced sample problem. Among them, w c Weights are assigned to corresponding categories, with rare anomaly categories receiving higher weights. After training, the model is lightweighted and deployed on an underground edge computing terminal or a ground monitoring host.

[0079] Online real-time identification: During construction, the system receives new time window data each time, and the model outputs the category probability vector of the current working condition in real time. If the "sealing failure" category probability is the highest and satisfies... If this occurs, a red alarm will be triggered, and suggestions for borehole sealing inspection and handling will be output simultaneously. This implementation method is particularly suitable for fracturing construction scenarios with complex geological conditions and blurred working boundary, and can significantly improve the accuracy of identifying hidden anomalies.

[0080] Example 3: Implementation Method for Effectiveness Level Classification and Construction Decision-Making This embodiment is applied to the effectiveness evaluation and construction decision-making of the target fracturing section in a coal mine. The specific implementation process is as follows: Real-time calculation of core indices: During construction, the system calculates the crack initiation confidence index I in real time. f =0.88, Extended Efficient Index I e =0.82, Abnormal Risk Index I a =0.18, Closing / Pressure Holding Reasonable Index I c =0.79.

[0081] Overall rating and grading: The weights are configured as follows: Calculate the overall score: According to the scoring range settings: Grade A: Score ≥ 0.85; Grade B: 0.70 ≤ Score < 0.85; Grade C: 0.55 ≤ Score < 0.70; Grade D: 0.40 ≤ Score < 0.55; Grade E: Score < 0.40. This section is judged as Grade A, indicating excellent construction results.

[0082] Linked construction decision-making: Based on the A-level effectiveness rating and the absence of abnormalities, the system outputs the following recommendations: continue injecting fluid at a stable pressure until the target injection volume is reached, then stop the pump and maintain the pressure. After the pressure maintenance is completed, proceed to the next stage of fracturing. The site followed these recommendations, and ultimately, the gas extraction concentration increased fourfold after this stage of fracturing, achieving the expected increased permeability.

[0083] A collaborative identification system for hydraulic fracturing construction anomalies and fracturing effectiveness is provided to implement the identification methods described in Examples 1-3. The system structure is as follows: Figure 3 As shown, it includes: The data acquisition module is used to collect geological data of the target fracturing reservoir, as well as multi-source real-time data during the fracturing operation, including sensor data such as pressure, displacement, and injection volume.

[0084] The preprocessing module communicates with the data acquisition module and is used to perform preprocessing such as time synchronization, filtering, and normalization on the acquired multi-source real-time data, and output a standardized input vector.

[0085] The stage identification module communicates with the preprocessing module and is used to divide the entire fracturing construction process into stages. Based on the standardized input vector, it completes the real-time identification of the current construction stage and outputs the stage state variable S(t).

[0086] The feature extraction module communicates with the preprocessing module and the stage identification module respectively, and is used to extract multi-source features corresponding to the current construction stage from the standardized input vector.

[0087] The index calculation module, which communicates with the feature extraction module, is used to construct and calculate the crack initiation confidence index I in real time based on the extracted multi-source features. f Extended Effective Index I e Abnormal Risk Index I a Three core quantitative indices.

[0088] The anomaly classification module communicates with the stage identification module and the index calculation module respectively. It is used to classify and identify the anomaly types in fracturing construction based on three types of core quantitative indices and stage state variables S(t), and output the anomaly type identification results.

[0089] The performance evaluation module communicates with the index calculation module and is used to combine three types of core quantitative indices with the reasonable closure / holding pressure index I. c Calculate the comprehensive score of fracturing effectiveness and classify the effectiveness level.

[0090] The handling suggestion linkage module communicates with the anomaly classification module and the effectiveness evaluation module respectively, and is used to output matching construction handling suggestions based on the anomaly type identification results and fracturing effectiveness level.

[0091] The alarm display module is connected to the anomaly classification module, the effectiveness evaluation module, and the handling suggestion linkage module. It is used to display the construction status and identification results in real time, and to issue audible and visual alarm prompts when an anomaly warning is triggered.

[0092] The storage module is connected to each of the above modules and is used to store raw data, preprocessed data, feature data, model parameters, identification results and processing suggestions.

[0093] This embodiment provides an electronic device, including a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it implements the method for collaborative identification of abnormalities and fracturing effectiveness in underground hydraulic fracturing construction in coal mines as described in any one of embodiments 1-3.

[0094] This embodiment also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the method for collaborative identification of abnormalities and fracturing effectiveness in underground hydraulic fracturing construction in coal mines as described in any one of embodiments 1-3.

[0095] The computer-readable storage medium can be a tangible device capable of holding and storing instructions used by an instruction execution device, such as, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. More specifically, the computer-readable storage medium includes: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, and any suitable combination thereof.

[0096] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for synergistic identification of hydraulic fracturing construction anomalies and fracturing effectiveness, characterized in that, Includes the following steps: S1. Data Acquisition and Standardized Preprocessing: Collect geological data of the target fracturing reservoir, and simultaneously collect multi-source real-time data during the fracturing operation. Preprocess the collected multi-source real-time data to obtain a standardized input vector. S2. Stage-by-stage state identification: The entire fracturing operation process is divided into at least five core stages. Based on the pre-processed standardized data, the current core stage of the operation is identified in real time, and stage state variables are generated. S3. Multi-source feature extraction in stages: For the current core construction stage identified in step S2, extract the multi-source features corresponding to the stage from the standardized input vector. S4. Collaborative Construction of Core Indices: Based on the multi-source features extracted in step S3, three types of core quantitative indices are constructed and calculated in real time, namely, the crack initiation confidence index, the expansion effectiveness index, and the anomaly risk index. S5. Classification of Construction Anomalies: Based on the three core quantitative indices obtained in step S4 and the current stage state variables obtained in step S2, the current fracturing construction conditions are classified and identified for anomalies, and the corresponding anomaly identification results are output. S6. Comprehensive evaluation of fracturing effectiveness: Based on the three core quantitative indices obtained in step S4, combined with the reasonable closure / pressure holding index, calculate the comprehensive fracturing effectiveness score of the current fracturing segment, and classify the fracturing effectiveness level according to the score results; S7. Action Recommendation Linkage Output: Based on the anomaly type identification results output in step S5 and the fracturing effectiveness level obtained in step S6, linkage output of construction action recommendations matching the current working conditions.

2. The method for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, In step S1, the multi-source real-time data during the fracturing operation includes at least: real-time orifice pressure data P(t), real-time fracturing operation discharge data Q(t), cumulative injection volume V(t), and real-time sand ratio sequence S. v (t), Real-time pressure data P of the fracturing section s (t), Pressure decay sequence P after pump shutdown sh (t), return liquid volume / return water volume sequence Q b (t), the water-seeping signal W(t) at the orifice; Preprocessing includes at least time synchronization, outlier removal, smoothing filtering, unified sampling interval reconstruction, working condition slicing, and stage labeling.

3. The method for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, In step S2, the real-time stage identification specifically includes the following steps: S21. Rule Prior Identification: The initial stage is determined based on prior rules of displacement change, pressure derivative, pump stoppage event, and injection volume increase. The specific determination rules are as follows: If real-time displacement And real-time pressure change rate If so, it is determined that the injection and pressurization phase has begun, in which... This refers to the injection threshold. If a local pressure peak occurs and is accompanied by a sudden drop in the pressure slope, it is determined that the crack initiation candidate stage has begun. If pumping continues and the pressure enters the fluctuation plateau range, it is determined that the crack has entered the crack propagation stage. If real-time displacement If the pressure decreases rapidly, it is determined that the pump has entered the pressure holding / closing response phase. If there is packer movement, tool string switching, or a continuous lack of fluid injection, it is determined that the inter-segment switching stage has begun; S22. Time series model correction: The initial stage identification results are smoothly corrected using existing methods. Unreasonable stage jumps are constrained by the state transition matrix, and the final stage state variables are output.

4. The method for collaborative identification of hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, In step S3, the specific calculation method for multi-source features is as follows: (1) Basic parameter characteristics: Including real-time pressure change rate Real-time displacement change rate Cumulative injection volume change rate Pressure derivative ; (2) Pressure-displacement coupling characteristics: including pressure increase coefficient per unit displacement Unit volume pressure increase coefficient Pressure fluctuation index ; (3) Pump shutdown attenuation characteristics: Fitting formula for pressure attenuation after pump shutdown ,in The instantaneous pressure at which the pump stops. To fit the closed characteristic pressure, The attenuation coefficient; (4) Internal and external pressure difference coefficient: sealing pressure difference ; (5) Crack propagation response characteristics: including propagation plateau stability Injection absorption index .

5. The method for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, In step S4, the crack initiation confidence index I f Used to depict the actual fracturing initiation state of the current fracturing section; when At that time, it was determined that crack initiation had been successful; when When the cracking is deemed insufficient or the initiation is uncertain, the following conditions are met: The threshold for determining crack initiation; Extended Effective Index I e Used to characterize the effective propagation state of cracks; when At that time, it was determined that the crack had effectively expanded, among which The threshold for determining the effective propagation of cracks; Abnormal Risk Index I a Used to characterize the abnormal risk level of the current fracturing operation; when When an anomaly warning is triggered, among which... This is the threshold for determining abnormal risks.

6. The method for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, In step S5, the abnormal types of fracturing conditions include at least: normal pressure initiation, successful fracturing, difficult fracturing / no fracturing, effective fracture propagation, insufficient fracture propagation / near-hole compaction, roadway leakage, excessive filtration loss, hole sealing failure / tool ​​string packer damage, cross-flow / channeling, abnormal pressure drop, normal pump shutdown closure, multiple fractures / fracture deflection, aquifer / water body perforation, abnormal gas, and abnormal mine pressure. The specific classification of the anomaly types is as follows: ① First, eliminate obviously normal and obviously abnormal operating conditions through mechanism rules; ② A machine learning model is used to classify complex boundary conditions in detail and output the results of anomaly type identification.

7. The method for collaborative identification of hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 5, characterized in that, In step S6, the expression for the comprehensive score of fracturing effectiveness is: in, Let be the weight coefficient, and satisfy... ; Based on the comprehensive scoring results, fracturing effectiveness is divided into five levels: A to E. ① Grade A: Score ≥ 0.85, indicating clear crack initiation, sufficient propagation, low risk of abnormalities, and excellent construction results; ② Grade B: 0.70≤Score<0.85, crack initiation is clear, and propagation is basically effective, with a slight risk; ③ Grade C: 0.55≤Score<0.70, cracks have started, but the expansion is moderate or there are local abnormalities, and the effect is moderate; ④ Grade D: 0.40≤Score<0.55, obvious construction abnormalities, poor effect in the current section; ⑤ Grade E: Score < 0.40, no cracking or severe failure, requires immediate action.

8. The method for collaboratively identifying hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, In step S7, the construction treatment recommendations include at least: ① continue pressure stabilization injection, ② reduce discharge rate to stabilize pressure, ③ increase discharge rate to promote fracturing, ④ temporarily stop pumping and repressurize, ⑤ check the sealing device and orifice seal, ⑥ identify suspected cross-sections and suspend construction, ⑦ end the current section and move to the next section, ⑧ supplement injection or repeat fracturing, ⑨ immediately stop construction if there is danger; a mapping relationship table between anomaly type, effectiveness level and treatment recommendations is established in advance, and the corresponding construction treatment recommendations are matched and output based on the anomaly type identification results of step S5 and the effectiveness level of step S6.

9. The method for collaborative identification of hydraulic fracturing construction anomalies and fracturing effectiveness according to claim 1, characterized in that, Step S5 further includes distinguishing between sealing failure and roadway leakage, specifically: Construct the discriminant function: in, For sealing pressure differential; This refers to the volume of returned liquid; For the pump shutdown pressure maintenance time, I f As the crack initiation confidence index, These are the weighting coefficients; when When the sealing fails, it is determined to be a failure; when At that time, it was determined to be a leak in the tunnel, among which To differentiate the judgment threshold; It also includes the identification of serial segments / streams, specifically: Construct segment risk items: in, For adjacent hole response intensity, To improve platform stability; , , For weighting coefficients; when When this occurs, a segment warning is triggered, in which... This is the threshold for segment determination.

10. A collaborative identification system for downhole hydraulic fracturing construction anomalies and fracturing effectiveness, characterized in that, The system for implementing the collaborative identification method according to any one of claims 1 to 9, the system comprising: The data acquisition module is used to collect geological data of the target fracturing reservoir, as well as multi-source real-time data during the fracturing operation. The preprocessing module, which communicates with the data acquisition module, is used to preprocess the acquired multi-source real-time data and output a standardized input vector. The stage identification module communicates with the preprocessing module and is used to divide the entire fracturing construction process into stages. Based on the standardized input vector, it completes the real-time identification of the current construction stage and outputs the stage state variables. The feature extraction module is connected to the preprocessing module and the stage identification module respectively. It is used to extract multi-source features of the corresponding stage from the standardized input vector for the identified current construction stage. The index calculation module communicates with the feature extraction module and is used to construct and calculate three core quantitative indices in real time: crack initiation confidence index, extended effective index, and abnormal risk index, based on the extracted multi-source features. The anomaly classification module communicates with the stage identification module and the index calculation module respectively. It is used to classify and identify the anomaly types in fracturing construction based on three types of core quantitative indices and stage state variables, and output the anomaly type identification results. The performance evaluation module communicates with the index calculation module and is used to calculate the comprehensive score of fracturing performance and classify the performance level based on three types of core quantitative indices combined with the reasonable closure / pressure holding index. The handling suggestion linkage module communicates with the anomaly classification module and the effectiveness evaluation module respectively, and is used to output matching construction handling suggestions based on the anomaly type identification results and fracturing effectiveness level. The alarm display module is connected to the anomaly classification module, the effectiveness evaluation module, and the handling suggestion linkage module, respectively. It is used to display the construction status and identification results in real time, and to issue alarm prompts when an anomaly warning is triggered. The storage module is connected to each of the above modules and is used to store raw data, preprocessed data, feature data, model parameters, identification results and processing suggestions.

Images