A fracturing channeling intelligent identification and dynamic regulation method based on multi-domain pressure derivative spectrum and multi-scale response characteristics

Abstract

This invention discloses an intelligent identification and dynamic control method for fracturing crosstalk based on multi-domain pressure derivative spectrum and multi-scale response characteristics, relating to the field of unconventional oil and gas field development technology. Addressing the shortcomings of traditional fracturing methods that rely solely on single-well pressure parameters, leading to false alarms and lacking spatial guidance, this method acquires high-frequency pressure data from multiple wells and constructs a multi-scale derivative spectrum to separate noise. It extracts the second-order catastrophe index characterizing pulse impact intensity, the high-frequency energy ratio characterizing physical channel type, and the spatial coordination coefficient depicting the dominant sweep direction from the time, frequency, and spatial domains, respectively. These multi-domain features are fused and projected onto an intelligent diagnostic chart, and threshold values ​​are used to accurately classify conditions such as sudden crosstalk in large channels, gradual sweep, and safe matrix seepage. Based on this, closed-loop control commands such as reducing discharge rate or temporarily plugging and redirecting the fracturing equipment are automatically triggered. This invention achieves millisecond-level quantitative early warning and automatic loss prevention for crosstalk, effectively ensuring the safe and efficient development of unconventional oil and gas multi-well platforms.

Description

Technical Field

[0001] This invention relates to the fields of unconventional oil and gas field development, hydraulic fracturing engineering, and artificial intelligence monitoring technology. Specifically, it relates to a real-time closed-loop control method that uses multi-domain pressure characteristics such as time-scale spectrum, second-order catastrophe index, multi-well collaboration, and frequency domain fingerprint to intelligently and quantitatively identify crosstalk between fracturing wells and dynamically adjust fracturing operation parameters. Background Technology

[0002] With the growth of global energy demand and the increasing depletion of conventional oil and gas resources, unconventional oil and gas resources such as shale gas and tight oil have become core areas for ensuring energy security. In the development of unconventional reservoirs, horizontal well networks combined with multi-stage large-scale hydraulic fracturing technology are key means to achieve economical and effective development. In order to further increase reservoir stimulation volume (SRV) and maximize single-well productivity, the industry has generally adopted the intensified stimulation strategy of "dense well network, long horizontal section, large displacement, high fluid volume, and large sand volume", as well as the operation mode of "multi-well platform zipper fracturing" [Zhao Jinzhou, Yong Rui, Hu Dongfeng, et al. Deep-ultra-deep shale gas fracturing in China: problems, challenges and development direction [J]. Acta Petrolei Sinica, 2024, 45 (1): 295-311]. Although this high-intensity operation mode has significantly improved the initial production, it has also brought about extremely complex inter-well interactions. As well spacing gradually decreases from the traditional 300-400 meters to 200 meters or even lower, the stress and pressure fields between adjacent wells frequently overlap, leading to increasingly serious inter-well fracturing crosstalk. According to relevant studies, in major shale oil and gas areas in North America and shale gas fields such as Changning-Weiyuan and Weirong in China, the incidence of fracturing crosstalk has reached as high as 50% to 70% [Li Yaqian, Yin Lishi, Li Zhili, et al. Research on the characteristics and control strategies of fracturing crosstalk in Weirong shale gas field [J]. Drilling and Production Technology, 2023, 46 (3)]. This phenomenon has evolved from an occasional engineering disturbance into a key bottleneck restricting the efficient development of unconventional oil and gas fields.

[0003] Inter-well fracturing crosstalk refers to the phenomenon where, during fracturing operations, artificial fractures or connecting natural fractures generated in the drilling well directly enter the drainage area of ​​an adjacent well (response well), or cause abnormal changes in the pressure and production of the adjacent well. From a physical mechanism perspective, crosstalk mainly includes two forms: pressure diffusion and direct fracture communication. This interference has a dual impact on gas field development, but in the current stage of dense well network development, the negative effects caused by fracturing crosstalk are becoming increasingly prominent, and have become a major factor affecting development efficiency: First, fracturing fluid enters the response well through crosstalk channels, which can lead to water lock damage, reservoir pressure loss, or a sudden increase in water cut in the response well. In the study of the Weirong shale gas field, it was found that severe crosstalk can lead to a significant decrease in the daily gas production of the response well, and subsequent recovery is difficult; Second, the high-pressure signal generated instantaneously by fracturing is rapidly transmitted to the wellbore of the response well through underground channels, which may cause wellhead pressure exceeding limits, tubing damage, or even blowout accidents, threatening the safety of construction personnel and equipment. Furthermore, energy leakage to adjacent wells means that the drilled well cannot form an effective fracture network within the designated area, resulting in insufficient stimulated volume (SRV) and severe ineffective pumping. Therefore, how to achieve accurate monitoring, quantitative identification, and real-time control of inter-well crosstalk has become an urgent international challenge to be solved in the field of smart fracturing for shale gas.

[0004] To identify and evaluate fracturing crosstalk, the industry currently mainly adopts three technical solutions: microseismic monitoring, fiber optic sensing (DAS / DTS), and adjacent well pressure monitoring. Microseismic monitoring technology: Although it can depict the spatial distribution of fractures in real time, it is limited by high cost and the signal-to-noise ratio requirements for signal identification in complex geological backgrounds, making it difficult to be widely adopted on all platforms [HOU B, ZHANG Q, LIU X, et al. Integration analysis of 3D fractures network reconstruction and frac hits response in shale wells[J]. Energy, 2022, 260:124906]. Fiber optic sensing technology: As a cutting-edge technology, fiber optics can capture minute vibrations and temperature changes, thereby identifying the approach of cracks [SAKAIDA S, PAKHOTINA I, ZHU D, et al. Evaluating effects of completion design on fracturing stimulation efficiency based on DAS and DTSinterpretation[C]. SPE Hydraulic Fracturing Technology Conference and Exhibition, The Woodlands, Texas, USA: SPE-209167-MS, 2022.]. However, fiber optic installation is extremely expensive and the downhole environment is complex, and its data interpretation often lags behind construction decisions. Therefore, the industry has begun to explore using low-cost and readily available adjacent well pressure changes for crosstalk interferometry analysis. Current mainstream methods include discrimination based on pressure amplitude changes and qualitative analysis based on the first or second derivative of pressure [CAI Y, DAHI TALEGHANI A. Using pressure changes in offset wells for interpreting fracture driven interactions (FDI)[J]. Journal of Petroleum Science and Engineering, 2022, 219: 111111.]. However, existing pressure-based monitoring and evaluation methods have serious shortcomings, mainly in the following aspects: First, they have poor noise resistance and lack multi-scale analytical capabilities. Traditional methods mostly rely on monitoring the absolute increase in pressure or simple first or second derivatives. However, under high-pressure, high-volume pumping conditions, the downhole pressure signal is accompanied by severe high-frequency water hammer noise, causing the derivative curve to oscillate violently, resulting in a large number of false alarms.Meanwhile, a single derivative cannot distinguish between "direct connection of the main fracture in a few seconds" and "micro-permeation in the matrix pores over tens of minutes," easily leading to false crosstalk misjudgments. This lack of multi-scale analytical capability makes it difficult for the system to accurately isolate effective crosstalk characteristic signals under complex pumping conditions. Second, there is a lack of automatically executable quantitative characterization indicators. Currently, judging derivative abrupt changes still largely relies on the visual interpretation of field engineers, lacking dimensionless quantitative indicators, and computers cannot automatically set thresholds for interception and judgment. Third, there are spatial limitations from a single-well perspective. Existing monitoring is mostly a point-to-point mode of "one construction well corresponding to one response well," ignoring that crosstalk is a three-dimensional spatial ripple process, and cannot determine the complexity and dominant trend of the fracture network through the synchronous response patterns of multiple adjacent wells. Fourth, the feature dimension is singular, ignoring frequency domain information. Existing analysis is entirely limited to the "time domain," failing to utilize "frequency domain" energy changes to distinguish between sudden large-channel water flow (dominated by high frequency) and slow micro-porous permeation (dominated by low frequency). Fifth, there is a lack of engineering closed-loop and real-time decision control. Current pressure monitoring systems are mostly open-loop "alarm systems" that cannot automatically generate intervention commands for frequency converters and fracturing pumps based on the physical mechanisms of crosstalk, and therefore cannot achieve intelligent loss prevention.

[0005] To date, existing inter-well fracturing crosstalk monitoring methods have significant shortcomings in identification accuracy, noise resistance, quantification, and real-time control capabilities, severely restricting the intelligentization of fracturing operations in unconventional oil and gas fields. The industry urgently needs a quantitative control method that can integrate multi-dimensional signal characteristics, possess strong anti-interference capabilities, and achieve a closed-loop flow of "monitoring-identification-decision-control." This invention aims to fill this technological gap by introducing techniques such as multi-scale signal processing, spatial vector recognition, and intelligent closed-loop control logic, achieving a leapfrog development from experience-based judgment to intelligent quantitative identification, and from open-loop monitoring to closed-loop automatic control. Summary of the Invention

[0006] This invention aims to overcome the shortcomings of existing technologies by proposing an intelligent identification and dynamic control method for fracturing crosstalk based on multi-domain pressure derivative spectrum and multi-scale response characteristics. This method considers the physical characteristics of fluid conduction at different time scales and establishes a comprehensive evaluation model based on multi-scale derivative spectrum, second-order catastrophe index, spatial synchronicity, and frequency domain fingerprint. Using this model, the crosstalk intensity can be quantified in real time and dynamic control strategies such as reducing discharge rate and temporary plugging of fracturing equipment can be automatically triggered.

[0007] The technical solution provided by this invention to solve the above-mentioned technical problems is: a method for intelligent identification and dynamic control of fracturing crosstalk based on multi-domain pressure derivative spectrum and multi-scale response characteristics, comprising the following steps: S1. Obtain high-frequency pressure data of the construction well and surrounding response wells, establish a baseline trend through a degradation algorithm, and extract the net pressure response quantity that characterizes the interference between wells; S2. Differentiate and integrate the net pressure response at different time scales to construct a multi-scale pressure derivative spectrum space, which is used to separate high-frequency noise and low-frequency seepage signal. S3. Extracting and defining the mutation index based on the second derivative ( I shock ), used to quantitatively characterize the sudden intensity of fluid disturbance at the moment of breakthrough; S4. Extract the derivative sequences of multiple response wells within the same time period and calculate the Pearson correlation coefficients between them. C ij ), used to characterize inter-well connectivity and dominant sweep direction; S5. Perform a Fourier transform on the pressure signal and calculate the ratio of high-frequency band energy to total energy. R high ), used for diagnosing burst-type crosstalk in the frequency domain; S6. Combine with the set multidimensional threshold To comprehensively determine whether interference has occurred and its type; S7. When certain high-risk triggering conditions are met, the system automatically performs closed-loop control to reduce discharge volume, suspend pumping, or add temporary plugging agent.

[0008] A further technical solution is that the specific processes of steps S1 and S2 are as follows: during the zipper-type fracturing operation, high-frequency monitoring pressure data of the response well is acquired in real time. p ( t The baseline trend is extracted using a low-pass filter. p ref ( t ), calculate the net pressure response. : ; Subsequently, its instantaneous first derivative was calculated. v ( t ) and second derivative a ( t ): ; Introducing time scale factor By performing sliding integral calculations at different scales, a derivative spectrum space is formed: ; in, Net pressure response, MPa; v ( t () is the instantaneous first derivative, MPa / min; a (t () is the instantaneous second derivative, MPa / min 2 ; For scale The multiscale derivative under the given condition is MPa / min.

[0009] A further technical solution is that the specific process of step S3 is as follows: To quantify the impulse characteristics of crosstalk and overcome the influence of baseline drift, the absolute value sequence of the second derivative is extracted within a set data window (containing N sampling points). a i |, define the mutation index as follows: ; in, I shock The abrupt change index is dimensionless; it quantitatively characterizes the pulse impact intensity when a fluid instantaneously breaks through a wall by calculating the ratio of the local maximum acceleration to the average acceleration.

[0010] A further technical solution is that the specific processes of steps S4 and S5 are as follows: Extract the first i Koujing and the first j Multiscale derivative sequence of wells v i and v j Calculate the Pearson correlation coefficient between the two: ; In the formula, C ij For well i With well j The coefficient of coordination between them, the closer the value is to 1, the stronger the spatial connectivity, and it is dimensionless; v i , v j These are the first-order pressure derivative sequences for the corresponding wells, in MPa / min; , This represents the mean of the corresponding sequence; Meanwhile, the net pressure signal in the time domain Apply a Fourier transform to map to the frequency domain: ; Calculate frequency domain characteristic indicators: ; in, The frequency domain complex spectrum function obtained after the time-domain pressure signal undergoes Fourier transform is expressed in MPa·min. Angular frequency, representing the rate of signal fluctuation, is expressed in rad / min. Represents the net pressure response signal in the time domain, in MPa; i The imaginary unit is dimensionless. t This represents the fracturing monitoring time, in minutes. R high The proportion of high-frequency energy is dimensionless. The preset high-frequency analysis band is in Hz; The highest cutoff frequency is Hz.

[0011] A further technical solution is that the specific processes of steps S6 and S7 are as follows: Establish a comprehensive judgment logic tree: when I shock > When, it is determined that crosstalk has occurred; when C ij > When, it is determined that a macroscopic connected path exists; when R high > At that time, it was determined to be a sudden type of intrusion; 1) If I shock > and C ij > and R high > If the area is identified as a sudden disturbance zone in the main channel, an automatic command to reduce the discharge rate (by 15%-30%) or an emergency pump shutdown will be issued until... I shock pullback; 2) If I shock ≤ but C ij > The area was determined to be in a gradual spillover equilibrium zone. The current discharge rate was maintained, and on-site application of a temporary plugging agent was recommended, along with... Figure 4 Topological relationship monitoring C ij Matrix changes were used to verify the steering effect; 3) If all indicators are below the threshold, it is determined to be a safe matrix seepage zone, the system will automatically filter and alarm, and maintain the original design parameters for stable construction.

[0012] The beneficial effects of this invention are: by pioneering the "multi-scale derivative spectrum," it effectively filters out high-frequency noise from fracturing pumping, accurately distinguishes instantaneous direct connection, delayed fracture propagation, and slow pore seepage from a time dimension, and solves the technical problem of false alarms easily caused by traditional single derivatives; it creatively proposes the second derivative mutation index (…). I shock ) and the energy ratio of high and low frequencies in the frequency domain ( E ratio This transforms the fuzzy curve shape into digital features, enabling computers to directly rely on safety thresholds for millisecond-level automatic interception; it also upgrades the system-level spatial perspective: the monitoring perspective is upgraded from a single well to a multi-well collaborative network, using a synchronous correlation matrix to quantitatively characterize the spatial propagation path of crosstalk and the complexity of the fracture network, providing three-dimensional guidance for complex fracture network transformation; it breaks the long-standing open-loop status quo of "only monitoring and not controlling", and establishes a complete automatic control logic tree from data acquisition and multi-dimensional discrimination to the operation of the fracturing truck frequency converter, realizing real-time automatic suppression and loss prevention of malignant crosstalk. Attached Figure Description

[0013] Figure 1 The method flowchart and system architecture closed-loop diagram provided for this invention; Figure 2 Construct comparative diagrams of derivative spectra at different time scales; Figure 3 This is a smart diagnostic chart for fracturing crosstalk based on multi-domain feature fusion; Figure 4 For multi-well spatial coordinated response C ij Topological relationship diagram for locating crosstalk direction. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further illustrated below with reference to the accompanying drawings and embodiments.

[0015] This invention provides an intelligent identification and dynamic control method for fracturing crosstalk based on multi-domain pressure derivative spectrum and multi-scale response characteristics. Its core lies in using multiple physical dimensions (time scale, non-Gaussian abrupt change, spatial coordination, frequency domain) to extract dimensionality-reduced features of inter-well interference phenomena and efficiently embedding them into a closed-loop control framework for real-time fracturing operations. The specific implementation steps are as follows (…). Figure 1 ): S1. Real-time data monitoring and net pressure signal extraction; S2, Discretization construction of multi-scale derivative spectrum space; S3. Transient impact quantification and mutation index calculation; S4. Spatial Synchronization Matrix and Cooperative Orientation Determination; S5. Time-frequency domain transformation and fingerprint feature extraction; S6. State machine-based comprehensive discrimination based on multi-dimensional thresholds; S7. Closed-loop execution of command issuance and equipment control.

[0016] In this invention, the specific data collection requirements for step S1 are as follows: during the zipper-type fracturing operation, combined with... Figure 1 The system architecture shown acquires real-time pressure data synchronously at a sampling frequency of not less than 1Hz by installing high-frequency pressure sensors on the excitation well W0 and at least two surrounding response wells W1, W2, ..., Wn. First, real-time construction parameters of the excitation well are acquired, including the construction displacement. Q Viscosity of fracturing fluid system and the total time of fracturing in the current stage T frac Simultaneously, a pre-defined baseline shunting algorithm is executed using the edge computing gateway to remove monitoring pressure. p ( t Static stratigraphic trend in ) p ref ( t ), calculate the net pressure response. : ; In the formula, Net pressure response, MPa; p ( t The real-time monitoring pressure of the response well is measured in MPa. p ref ( t () is the reference pressure, MPa.

[0017] In this invention, the specific process of step S2 is as follows: the system performs multi-timescale analysis on the acquired net pressure response; and sets integration windows at the second level (short scale), minute level (medium scale), and stage level (long scale) respectively. Multi-scale derivatives are constructed by performing a sliding weighted integral on the rate of pressure change at various scales. , forming as Figure 2 The derivative spectrum space shown is used to physically separate high-frequency noise from low-frequency percolation signals: ; In the formula, In order to scale Pressure derivative spectrum value, MPa / min; This represents the first-order rate of change of pressure over time. denoted as min, representing the integration time window length.

[0018] In this invention, the specific process of step S3 is as follows: the system calculates the second-order time derivative of the pressure signal in real time. a ( t The acceleration of the pressure response is used to characterize the acceleration. Within a set sampling data window of N, the maximum extreme value and arithmetic mean of the absolute value of the second derivative are extracted to calculate the dimensionless catastrophe index. I shock To quantitatively characterize the sudden impact intensity of fluids ( Figure 3 (horizontal axis) ; In the formula, I shock The mutation index is dimensionless. a i For the first i Second derivative of pressure at each sampling point, MPa / min 2 N represents the total number of samples within the sliding calculation window.

[0019] In this invention, the specific process of step S4 is as follows: synchronously extracting the first derivative sequence of adjacent multi-well response. v i and v j The spatial co-response matrix is ​​constructed in real time using Pearson correlation analysis, and the correlation coefficient is calculated. C ij ( Figure 4 This is used to determine the dominant propagation path of interference in the underground network: ; In the formula, C ij For well i With well j The co-response coefficient between them is dimensionless; v i , v j These are the first-order pressure derivative sequences for the corresponding wells, in MPa / min; , This represents the mean of the corresponding sequence.

[0020] In this invention, step S5 specifically involves: performing a fast Fourier transform on the pressure fluctuation signal to map the time-domain signal to the frequency domain; and calculating the ratio of high-frequency band (representing fluid hammer) energy to total energy through energy integration. R high Fingerprint identification as a physical channel type ( Figure 3 (Vertical axis) ; ; In the formula, The frequency domain complex spectrum function obtained after the time-domain pressure signal undergoes Fourier transform is expressed in MPa·min. Angular frequency, representing the rate of signal fluctuation, is expressed in rad / min. Represents the net pressure response signal in the time domain, in MPa; i The imaginary unit is dimensionless. t This represents the fracturing monitoring time, in minutes. R high The proportion of high-frequency energy is dimensionless. The preset high-frequency analysis band is in Hz; The highest cutoff frequency is Hz.

[0021] In this invention, the specific process of step S6 is as follows: the system calculates the above-mentioned... I shock , C ij , R high The system inputs multi-domain feature quantities into a pre-defined intelligent discrimination state machine; this state machine is based on multi-dimensional thresholds pre-calibrated according to the geological characteristics of the block. Perform hierarchical logical deduction to determine the risk level of the current interference in real time, and combine it with Figure 3 Logical discrimination is performed on the discriminant space shown: 1) If I shock > and C ij > and R high > It was determined to be a sudden disturbance zone on the main channel (State A). 2) If I shock ≤ but C ij > It is determined to be a gradual ripple equilibrium zone (state B). 3) If all indicators are below the threshold, it is determined to be a safe matrix seepage zone (state C).

[0022] In this invention, the specific process of step S7 is as follows: based on the discrimination result of step S6, according to... Figure 1 The closed-loop system, through which the control system directly sends commands to the fracturing actuator via an industrial communication protocol, allows the fracturing actuator to operate. 1) Trigger State A: Automatically issues a displacement reduction command (reduction range 15%-30%) or emergency pump stop, until... I shockpullback; 2) Trigger Status B: Maintain current displacement, prompt on-site addition of temporary plugging agent, and combine with Figure 4 Topological relationship monitoring C ij Matrix changes were used to verify the steering effect; 3) Triggering state C: The system automatically filters alarms and maintains stable construction with the original design parameters.

[0023] Example 1: A deep shale gas multi-well platform (fractured well W0, offset response wells W1, W2, and W3), the field data acquisition system uses... f Pressure data is recorded in real time at a frequency of 1Hz (i.e., 1 sampling point per second). The system's preset safety thresholds are: mutation index threshold. =5.0, spatial connectivity threshold =0.6, frequency domain energy threshold =0.20.

[0024] 1. Quantitative identification and closed-loop interception of sudden strong crosstalk in large channels (State A): (1) Basic data acquisition (step S1): Fracturing the W0 well to the 8th stage (displacement rate) Q =16m 3 When / min), the bottom hole pressure of response well W1 p ( t )exist t A sharp fluctuation occurred starting at 15 minutes; high-frequency pressure data before and after this period were extracted, and after low-pass filtering to remove static pressure drift, the baseline pressure was measured. p ref ( t =40MPa; in the subsequent short time window Within 10 minutes (i.e., 1 / 6 minute), the pressure in well W1 rapidly increased to 42.5 MPa. The net pressure response at this point is calculated as follows: =2.5MPa; (2) Calculation of derivative spectrum and acceleration (step S2): System computation of short scale ( The first derivative of the integral (velocity) at (=1 / 6min) is: ; Meanwhile, the system also processed the data points within the last 10 seconds (a total of 10 points, denoted as...). N =10) Perform a second difference to obtain the instantaneous second derivative (acceleration) sequence. a i Calculations show that the maximum acceleration extreme value max(|) caused by fluid impact within this time window is... a i |)=4.5MPa / min2 The average acceleration background value within that window .

[0025] (3) Mutation intensity quantification (step S3): Substituting the above second derivative parameters into the mutation index formula, calculate... I shock : ; determination: I shock (18)> (5) The system determines that an abnormally high fluid pulse impact has occurred.

[0026] (4) Spatial connectivity and frequency domain discrimination (steps S4 and S5): The first derivative sequence of well W2 was extracted simultaneously during the same period. After covariance calculation, the covariance between the derivative sequences of W1 and W2 was 4.15, and their standard deviations were 2.1 and 2.3, respectively. Substituting these values ​​into the Pearson formula, the spatial covariance coefficient was calculated to be 0.86. Meanwhile, a Fourier transform was performed on the net pressure signal of well W1, and the total energy across the entire frequency band was found to be 1500J, of which the energy of high-frequency water hammer waves greater than 1Hz was 630J, and the proportion of high-frequency energy was 0.42%. (5) Comprehensive judgment and action (steps S6-S7): Due to I shock >5.0, C 12 > 0.6 and R high >0.2, the state machine confirms the occurrence of "sudden crosstalk in a large channel with a clear dominant direction"; the system immediately sends a command to the W0 well frequency converter within 3 seconds to automatically reduce the displacement by 20% (to 12.8m). 3 / min), forcibly weakening the destructive power of the water impact.

[0027] 2. Intelligent identification and steering control of progressively affecting complex fracture networks (State B): When well W0 is fractured to the 12th stage, the pressure in response well W2 remains constant for a long period of time. Within a mesoscale time window of 5 minutes, the pressure slowly and smoothly increased from 40 MPa to 43.5 MPa. The mesoscale first derivative is calculated to be 0.7 MPa / min. The second derivative sequence within this 5-minute period (N=300 sampling points) is extracted, and the maximum value (| a i |)=0.56MPa / min 2 ,average value The calculated mutation index is 3.5; the calculated frequency domain energy ratio (low-frequency deformation energy dominates) is 0.12; the spatial coordination matrix shows that the derivative coordination coefficients of wells W2 and W3 are... C 23 =0.88; although C 23 (0.88)> (0.6) indicates that the fluid spreads to the W2 / W3 region, but I shock (3.5) (5.0) and R high (0.12)< The system determined that there was no malignant large-channel crosstalk, but rather a gradual ripple effect of the micro-suturing network in a balanced state. The system did not trigger a reduction in discharge rate, maintaining full-load operation, but instead output a command to the control panel stating, "Micro-suturing network development detected in the W2-W3 direction; immediate application of temporary plugging agent recommended." After 150 kg of temporary plugging agent was applied on-site, the system monitored the mesoscale derivative. The descent begins, confirming successful internal turning.

[0028] Intelligent immune system to prevent false alarms related to safe matrix seepage (Status C): During high-volume pumping of well W0, the pressure in well W3 400 meters away is... Within a long-scale time window of 30 minutes, the pressure gradually increased from 35 MPa to 35.5 MPa. If the traditional method is used (pressure increase exceeding 0.5 MPa), the system is highly prone to triggering false alarms; the long-scale derivative is 0.016 MPa / min; the extreme value and mean of the second derivative are almost equal during this period, which is calculated to be... I shock =1.05; Frequency domain analysis shows that the signal energy is extremely low and entirely low frequency ( R high =0.02); multi-well correlation calculations show that W3 correlates with any well. C ij All values ​​were less than 0.2; all characteristic parameters were far below the warning threshold, and the system determined that the pressure rise was a safe formation pore elastic deformation and matrix seepage; the system intelligently filtered out the fluctuation signal, without triggering any audible or visual alarms, ensuring the continuity of fracturing operations and the efficiency of reservoir stimulation.

[0029] Based on the intelligent identification and dynamic control method described in this invention, firstly, by deeply integrating multi-scale derivative spectra, second-order catastrophe exponents, multi-well spatial synergy matrices, and time-frequency domain energy fingerprints, the method accurately characterizes the multi-dimensional evolutionary features of transient hydraulic impact, fracture network progressive waves, and matrix pore seepage during fracturing crosstalk, providing solid data and theoretical support for quantitatively assessing the physical mechanisms of crosstalk and the dominant direction of fracture propagation. Secondly, it innovatively constructs a multi-dimensional intelligent early warning state machine spanning the "time domain-frequency domain-space," completely overcoming the limitations of traditional fracturing methods that rely solely on time-frequency-space data. The invention overcomes the technical bottlenecks caused by relying on single pressure amplitude or simple derivatives, which lead to "high-frequency noise easily causing misjudgments" and "lack of global guidance from a single well perspective." Ultimately, this invention not only achieves a historic leap from "one-dimensional qualitative post-hoc analysis" to "multi-domain quantitative millisecond-level perception" in the evaluation of inter-well crosstalk, but also provides highly forward-looking and engineering-application-valued technical guidance for the safe and efficient development of unconventional oil and gas multi-well platforms, the sub-minute-level intelligent closed-loop control of fracturing trucks, and the maximization of reservoir stimulation volume (SRV) and final single-well recovery rate (EUR).

[0030] The above description is not intended to limit the present invention in any way. Although the present invention has been disclosed through the above embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for intelligent identification and dynamic control of fracturing crosstalk based on multi-domain pressure derivative spectrum and multi-scale response characteristics, characterized in that, Includes the following steps: S1. Obtain high-frequency pressure data of the construction well and surrounding response wells, establish a baseline trend through a degradation algorithm, and extract the net pressure response quantity that characterizes the interference between wells; S2. Differentiate and integrate the net pressure response at different time scales to construct a multi-scale pressure derivative spectrum space, which is used to separate high-frequency noise and low-frequency seepage signal. S3. Extracting and defining the mutation index based on the second derivative ( I shock ), used to quantitatively characterize the sudden intensity of fluid disturbance at the moment of breakthrough; S4, extract the derivative sequence of multi-responding wells in the same period, calculate the Pearson correlation coefficient between each other (R) C ij ), used to describe interwell connectivity and dominant sweep direction; S5. Fourier transform of the pressure signal, calculation of the ratio of the high frequency band energy to the total energy R high for diagnosing burst-type crosstalk from the frequency domain; S6, combine the set multi-dimensional threshold values , comprehensive discrimination interference whether it occurs and its type; S7. When certain high-risk triggering conditions are met, the system automatically performs closed-loop control to reduce discharge volume, suspend pumping, or add temporary plugging agent.

2. The intelligent identification and dynamic control method for fracturing crosstalk based on multi-domain pressure derivative spectrum and multi-scale response characteristics according to claim 1, characterized in that, The specific process of step S1 is as follows: First, obtain the real-time construction parameters of the excitation well, including the construction discharge rate. Q Viscosity of fracturing fluid system and total fracturing time in the current stage T frac Simultaneously, the edge computing gateway executes a pre-defined baseline fading algorithm to remove monitoring pressure. p ( t Static stratigraphic trend in ) p ref ( t ), calculate the net pressure response: ； In the formula, Net pressure response, MPa; p t ) as a response to well real-time monitoring pressure, MPa;​ p ref ( t ) is the reference pressure, MPa.

3. The method of claim 1, wherein, The specific steps of S2 include: The system performs multi-timescale analysis on the acquired net pressure response, setting integration windows at the second (short-scale), minute (medium-scale), and stage (long-scale) levels respectively. Multi-scale derivatives are constructed by performing a sliding weighted integral on the rate of pressure change at various scales. This forms a derivative spectrum space, used for physically separating high-frequency noise from low-frequency percolation signals. ； wherein is the pressure derivative spectrum value at the scale MPa / min; is the first order rate of change of pressure with time; is the integration time window length, min.

4. The method of claim 1, wherein, The specific steps of S3 include: The system calculates the second-order time derivative of the pressure signal in real time. a ( t The acceleration of the pressure response is used to characterize the acceleration. Within a set sampling data window of N, the maximum extreme value and arithmetic mean of the absolute value of the second derivative are extracted to calculate the dimensionless catastrophe index. I shock To quantitatively characterize the intensity of sudden fluid impact: ； In the formula, I shock The mutation index is dimensionless. a i For the first i Second derivative of pressure at each sampling point, MPa / min 2 N represents the total number of samples within the sliding calculation window.

5. The method of claim 1, wherein, The specific steps of S4 include: Synchronously extracting first derivative sequences of adjacent multi-response wells v i With v j ; real-time construction of spatially cooperative response matrix by Pearson correlation analysis, calculation of correlation coefficient C ij , to identify the dominant propagation path of channeling in the underground fracture network ； In the formula, C ij For well i With well j The co-response coefficient between them is dimensionless; v i , v j These are the first-order pressure derivative sequences for the corresponding wells, in MPa / min; , This represents the mean of the corresponding sequence.

6. The method of claim 1, wherein, The specific steps of step S5 include: A fast Fourier transform is performed on the pressure fluctuation signal to map the time-domain signal to the frequency domain; the ratio of high-frequency band energy (representing fluid hammer) to total energy is calculated by energy integration. R high Fingerprint identification as a physical channel type: ； ； In the formula, The frequency domain complex spectrum function obtained after the time-domain pressure signal undergoes Fourier transform is expressed in MPa·min. Angular frequency, representing the rate of signal fluctuation, is expressed in rad / min. Represents the net pressure response signal in the time domain, in MPa; i The imaginary unit is dimensionless. t This represents the fracturing monitoring time, in minutes. R high The proportion of high-frequency energy is dimensionless. The preset high-frequency analysis band is in Hz; The highest cutoff frequency is Hz.

7. The method of claim 1, wherein, The specific steps of step S6 include: The system will use the above calculations to obtain I shock , C ij , R high The system inputs multi-domain feature quantities into a pre-defined intelligent discrimination state machine; this state machine is based on multi-dimensional thresholds pre-calibrated according to the geological characteristics of the block. Hierarchical logical derivation is performed to determine the risk level of the current crosstalk in real time, and logical discrimination is made by combining the intelligent diagnostic chart for fracturing crosstalk with multi-domain feature fusion. 1) if I shock > and C ij > and R high > , a large channel burst crosstalk region (state A) is determined. 2) if I shock ≤ But C ij > , the progressive wave and the equilibrium region (state B) are determined. 3) If all indicators are below the threshold, it is determined to be a safe matrix seepage zone (state C).

8. The method of claim 1, wherein, The specific steps of step S7 include: Based on the judgment result of step S6, and according to the closed-loop system in Figure 1, the control system directly sends commands to the fracturing actuator through the industrial communication protocol: 1) Trigger State A: Automatically issues a displacement reduction command (reduction range 15%-30%) or emergency pump stop, until... I shock pullback; 2) Triggering State B: Maintain the current discharge rate, prompt the field to add temporary plugging agent, and combine multi-well spatial collaborative response with topological relationship monitoring of crosstalk direction. C ij Matrix changes were used to verify the steering effect; 3) Triggering state C: The system automatically filters alarms and maintains stable construction with the original design parameters.

Images