Gas storage safety monitoring method based on well-ground joint observation optimization and related equipment

Abstract

The application provides a kind of gas storage safety monitoring method based on well-ground joint observation optimization, first, the physical field monitoring data of target gas storage is collected according to the three-dimensional observation network constructed at the target gas storage;Microseismic event detection and three-dimensional joint positioning are carried out on the physical field monitoring data, and the microseismic event positioning result is obtained;Based on the microseismic event positioning result and the physical field monitoring data, the inversion of seismic source mechanism and the multi-scale inversion of regional stress field are carried out, and the mechanical parameter inversion result is obtained;Based on the microseismic event positioning result, the mechanical parameter inversion result and the physical field monitoring data, dynamic risk assessment and hierarchical early warning are carried out.The application realizes the whole-chain intelligent monitoring of gas storage from data perception to mechanical diagnosis to decision-making and early warning, and realizes the safety monitoring demand of high precision, high stability and full-process automation of gas storage.

Description

Technical Field

[0001] This invention relates to the field of gas storage engineering safety and geophysical exploration monitoring technology, specifically to a gas storage safety monitoring method and related equipment based on well-ground joint observation optimization. Background Technology

[0002] Underground gas storage facilities are core infrastructure for ensuring the security of the natural gas supply chain, achieving seasonal peak shaving, and establishing strategic reserves. Their importance is increasingly evident as the global energy transition deepens. During long-term cyclic injection and production operations in natural gas storage facilities, the surrounding rock and overlying strata of the storage chamber are subjected to significant periodic alternating loads. This process can induce a series of rock mechanical responses, including salt rock creep, fracture initiation and propagation, fault activation, and even perceptible earthquakes. These physical processes are typically accompanied by the release of transient elastic waves from microfractures, i.e., microseismic signals. By monitoring and analyzing these microseismic signals, the damage evolution mechanisms and potential instability modes of the surrounding rock and caprock of the gas storage facility can be effectively revealed.

[0003] However, existing research on microseismic monitoring in gas storage facilities mainly focuses on parameter analysis such as event counting, energy analysis, and magnitude statistics, while research on the spatial location of microseismic events is relatively insufficient. Furthermore, existing microseismic monitoring systems for gas storage facilities generally suffer from the following problems: ① Sensor (detector) deployment relies heavily on empirical design, lacking systematic optimization based on the specific geological structure and expected rupture mechanisms of the gas storage facility, easily leading to incomplete monitoring coverage and insufficient positioning accuracy in key risk areas; ② The acquisition and positioning calculation of microseismic signals are easily affected by environmental noise and repetitive noise from injection and production operations, resulting in poor stability and serious missed detection of low signal-to-noise ratio events, making it difficult to simultaneously achieve computational efficiency and calculation accuracy; ③ Monitoring remains at the stage of phenomenon description, lacking the ability to diagnose rupture driving mechanisms and predict trends; ④ There is a lack of full life-cycle dynamic monitoring and multi-source data fusion evaluation methods, and the real-time performance and intelligence level need to be improved.

[0004] In other words, how to provide a gas storage safety monitoring method based on well-ground joint observation optimization, which can meet the safety monitoring requirements of gas storage with high precision, high stability and full-process automation, and achieve effective coverage of key risk areas and accurate diagnosis of rupture mechanisms, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] This invention provides a gas storage safety monitoring method and related equipment based on well-ground joint observation optimization to solve at least one of the above-mentioned technical problems.

[0006] Firstly, this application provides a gas storage safety monitoring method based on well-ground joint observation optimization, the method comprising: Physical field monitoring data of the target gas storage facility are collected based on a three-dimensional observation network constructed at the target gas storage facility. Microseismic event detection and three-dimensional joint localization are performed on the physical field monitoring data to obtain microseismic event localization results; Based on the microseismic event location results and the physical field monitoring data, source mechanism inversion and regional stress field multi-scale inversion were carried out to obtain mechanical parameter inversion results. Dynamic risk assessment and graded early warning are conducted based on the microseismic event location results, the mechanical parameter inversion results, and the physical field monitoring data.

[0007] Preferably, the three-dimensional observation network consists of an air-assisted monitoring unit, a ground monitoring unit, and a downhole monitoring unit; the air-assisted monitoring unit uses satellite InSAR and UAV survey equipment to conduct periodic surface monitoring; the ground monitoring unit includes several ground stations, and the downhole monitoring unit includes several monitoring wells and downhole sensor strings; the ground stations, monitoring wells, and downhole sensor strings are deployed according to a well-ground joint observation optimization scheme.

[0008] Preferably, the well-to-surface joint observation optimization scheme is obtained through the following steps: The three-dimensional geomechanical model of the target gas storage facility, the lower limit of the target monitoring magnitude, the preset risk area, and the well-ground engineering constraints are obtained. The well-ground engineering constraints include the upper limit of the number of surface stations and monitoring wells, the upper limit of the number of downhole sensors, the wellhead location limit, the cable length limit, and the deployment cost constraint. Based on the aforementioned three-dimensional geomechanical model, the lower limit of the target monitoring magnitude, and the preset risk area, a multi-objective optimization function is constructed with positioning accuracy and spatial coverage as the core. The optimization objective of positioning accuracy is to maximize the determinant value of the Fisher information matrix, and the optimization objective of spatial coverage is to maximize the theoretically detectable spatial coverage. A multi-objective optimization problem is constructed based on the multi-objective optimization function and well-ground engineering constraints. The multi-objective optimization problem is solved iteratively using a non-dominated sorting genetic algorithm with an elite strategy to obtain a Pareto optimal solution set. Based on actual engineering requirements, the optimal implementation scheme is selected from the Pareto optimal solution set to obtain the well-ground joint observation optimization scheme; The well-ground joint observation optimization scheme includes at least the latitude and longitude coordinates and layout of the ground stations, the location coordinates and number of monitoring wells, and the depth range and level spacing of the downhole sensor strings; the layout can be either a regular grid or a surrounding cavity.

[0009] Preferably, the physical field monitoring data includes continuous waveform data, and the specific process of detecting microseismic events using the physical field monitoring data includes: Continuous waveform data is entered into a message queue and sliced ​​according to a fixed duration to obtain waveform slice data. The waveform slice data is input into the trained improved U-Net convolutional neural network model to obtain the probability values ​​of P-wave, S-wave and noise output point by point along the time axis, which constitute a time-series probability sequence. Suspected microseismic event segments are extracted based on the time-series probability sequence. The suspected microseismic event fragment and its corresponding original continuous waveform data are input into the Transformer encoder. The authenticity of the event is determined by the self-attention mechanism to obtain the real microseismic event. The P-wave and S-wave first arrival times of the real microseismic event are picked up to obtain the P / S-wave first arrival times of each real microseismic event.

[0010] Preferably, the specific process of the three-dimensional joint positioning includes: Based on the first arrival times of the P-wave and S-wave of each real microseismic event, the simplex method is used to locate each real microseismic event and obtain preliminary location results. Based on the waveform cross-correlation coefficients between the real microseismic events, the real microseismic events are associated and paired to obtain multiple event pairs; Construct double-difference localization equations for all event pairs and solve them to obtain the location correction parameters for each of the real microseismic events. The preliminary location results of the actual microseismic event are corrected by the location correction parameters to obtain the microseismic event location results.

[0011] Preferably, the double-difference localization equation for each pair of events is: ; in, For events in an event pair and events Reach any observation point The travel time difference of P / S waves, where the observation point is any one of the ground station and the monitoring well; For the event Arrive at the observation point The travel time of the P / S wave; For the event Arrive at the observation point P / S wave travel time; It is an event and events Spatial location difference, It is an event and events The time difference of the earthquake's occurrence; For the event The travel-time partial derivative; For the event The traveling partial derivatives.

[0012] Preferably, the specific process of the focal mechanism inversion includes: Based on the microseismic event location results, valid microseismic events with a sufficient number of effective records at observation points and complete waveform data were selected. Based on the location results of the effective microseismic events, six sets of basic Green's functions corresponding to each observation point of the effective microseismic event are matched from a pre-constructed three-dimensional Green's function database. The six sets of basic Green's functions correspond one-to-one with the six independent components of the moment tensor. For each effective microseismic event, the theoretical simulated waveform is obtained by convolving each set of basic Green's functions with the corresponding moment tensor components. With the goal of minimizing the L2 norm residual between the measured waveform and the theoretical simulated waveform, the optimal moment tensor for each effective microseismic event is obtained by iterative optimization using the adjoint state method combined with the L-BFGS algorithm. The optimal moment tensor is orthogonally decomposed to obtain the source rupture characteristic parameters of each effective microseismic event; The three-dimensional Green's function database is constructed according to the following steps: the study area of ​​the three-dimensional geomechanical model is divided into grids according to space and depth to obtain multiple grid nodes; based on the three-dimensional velocity model, the spectral element method is used to calculate the six sets of basic Green's functions corresponding to each observation point of each grid node; and the Green's functions of all grid nodes are collected to form a three-dimensional Green's function database covering the study area.

[0013] Preferably, the specific process of the multi-scale inversion of the regional stress field includes: Based on the microseismic event location results, the effective microseismic events are divided into three time series subsets: the pre-pressurization stage, the rapid pressurization stage, and the pressure stabilization stage. Based on the Wallace-Bott mechanical assumptions, and constrained by the source rupture characteristic parameters of microseismic events within each time series subset, the stress tensor of the corresponding region in the corresponding time period is inverted using the Bayesian MCMC method to obtain the stress field inversion results for each time series stage. The mechanical parameter inversion results include the stress field inversion results and the source rupture characteristic parameters.

[0014] Secondly, this application also provides a gas storage safety monitoring device optimized based on well-ground joint observation, the device comprising: The signal acquisition unit is used to acquire physical field monitoring data of the target gas storage facility based on the three-dimensional observation network constructed at the target gas storage facility. The detection and positioning unit is used to perform microseismic event detection and three-dimensional joint positioning on the physical field monitoring data to obtain the microseismic event positioning results. The data inversion unit is used to perform source mechanism inversion and regional stress field multi-scale inversion based on the microseismic event location results and the physical field monitoring data, and to obtain mechanical parameter inversion results. The assessment and early warning unit is used to perform dynamic risk assessment and graded early warning based on the microseismic event location results, the mechanical parameter inversion results, and the physical field monitoring data.

[0015] Thirdly, this application also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to execute the computer program stored in the memory to implement any of the gas storage safety monitoring methods based on well-ground joint observation optimization in the first aspect.

[0016] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any one of the gas storage safety monitoring methods based on well-ground joint observation optimization in the first aspect.

[0017] This invention first collects physical field monitoring data of the target gas storage facility based on a three-dimensional observation network constructed at the target gas storage facility; then performs microseismic event detection and three-dimensional joint localization on the physical field monitoring data to obtain microseismic event localization results; based on the microseismic event localization results and physical field monitoring data, it conducts source mechanism inversion and regional stress field multi-scale inversion to obtain mechanical parameter inversion results; and finally, based on the microseismic event localization results, mechanical parameter inversion results, and physical field monitoring data, it conducts dynamic risk assessment and graded early warning. This application improves monitoring sensitivity and positioning accuracy from the source by constructing a three-dimensional observation network; it achieves unattended automated processing through event detection and picking, minimizing human intervention; it provides rupture mechanism and stress state information that traditional monitoring cannot obtain through full-waveform moment tensor inversion and stress field inversion, providing a solid physical and mechanical foundation for risk assessment; through the complete process of three-dimensional observation network acquisition, intelligent detection and positioning, multi-scale mechanical inversion and multi-parameter risk early warning, it realizes intelligent monitoring of the entire chain of gas storage facilities from data perception to mechanical diagnosis to decision-making and early warning, achieving the safety monitoring requirements of high precision, high stability and full-process automation for gas storage facilities, as well as effective coverage of key risk areas and accurate diagnosis of rupture mechanisms. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic diagram of a gas storage safety monitoring method based on well-ground joint observation optimization provided in this application; Figure 2 The overall workflow diagram of the safety monitoring method for the target gas storage facility provided in this application; Figure 3 This is a schematic diagram of the combined surface and downhole observation network provided in this application; Figure 4 A schematic diagram of the cascaded microseismic event intelligent detection and first arrival picking process provided in this application; Figure 5 A schematic diagram illustrating the principle of the waveform cross-correlation-based double-difference localization method provided in this application; Figure 6 This is a structural diagram of a salt cavern gas storage facility provided in this application; Figure 7 This is a schematic diagram of the deployment and data transmission of the three-dimensional monitoring network provided in this application; Figure 8 A schematic diagram of a gas storage safety monitoring device based on well-ground joint observation optimization provided in this application; Figure 9 A schematic diagram of the structure of the electronic device provided in this application; Figure 10 A schematic diagram of a structure of a computer-readable storage medium provided in this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules is not necessarily limited to those explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices. The naming or numbering of steps appearing in this application does not imply that the steps in the method flow must be performed in the chronological / logical order indicated by the naming or numbering. The execution order of named or numbered process steps can be changed according to the desired technical purpose, as long as the same or similar technical effect is achieved.

[0022] The module division described in this application is a logical division. In practical applications, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection between modules shown or discussed may be through some interfaces, and the indirect coupling or communication connection between modules may be electrical or other similar forms, none of which are limited in this application. Furthermore, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed in multiple circuit modules. Some or all of the modules may be selected to achieve the purpose of the solution in this application according to actual needs.

[0023] Next, please refer to Figure 1-5 , Figure 1 This is a flowchart illustrating a gas storage safety monitoring method based on well-to-surface joint observation optimization, as an embodiment of the present invention. As an example of the gas storage safety monitoring method based on well-to-surface joint observation optimization provided by the present invention, the method specifically includes the following steps S110 to S140: Step S110: Collect physical field monitoring data of the target gas storage facility based on the three-dimensional observation network constructed at the target gas storage facility; As one feasible approach, the three-dimensional observation network in step S110 specifically consists of an airborne auxiliary monitoring unit, a ground monitoring unit, and a downhole monitoring unit. The airborne auxiliary monitoring unit uses satellite InSAR and UAV survey equipment for periodic surface monitoring. The ground monitoring unit includes several ground stations, and the downhole monitoring unit includes several monitoring wells and downhole sensor strings. The ground stations, monitoring wells, and downhole sensor strings are deployed according to the well-ground joint observation optimization scheme. An example of the network deployment structure for joint ground and downhole observation is shown below. Figure 3 .

[0024] Based on the design scheme of step S110, a three-dimensional observation network of "star (InSAR) - air (UAV survey) - ground (earthquake, tilt, GPS) - well (earthquake, acoustic wave, strain, pressure)" is constructed.

[0025] The ground monitoring unit consists of several unattended intelligent observation stations. Each station includes: a weatherproof enclosure, a three-component seismometer (frequency band 0.1 Hz-500 Hz), a dual-axis tiltmeter, a GNSS antenna and receiver, a main control / acquisition unit (including a high-precision clock, large-capacity storage, and edge computing unit), a power system (solar / battery / mains power), and communication equipment (4G / 5G / fiber optic transceivers). The seismometer is used to capture medium- and high-frequency microseismic signals; the tiltmeter is sensitive to large-area, slow deformation; and the GNSS provides an absolute displacement reference.

[0026] Downhole monitoring unit: Includes one or more dedicated monitoring wells. Installed inside the wellbore: Multi-stage, three-component downhole geophone strings (stage spacing 10-50 meters), encapsulated in a protective housing; Parallel to these are distributed fiber optic acoustic wave sensor (DAS) or fiber Bragg grating (FBG) sensor strings for continuous distributed strain / temperature measurement; Wellhead equipment, data acquisition cabinet, and optical transceiver are installed at the wellhead. The DAS / FBG provides high spatial resolution continuous strain / temperature profiles.

[0027] Aerial auxiliary monitoring unit: Satellite InSAR data interpretation is activated periodically or as needed, and UAV-borne thermal infrared or lidar is used to conduct surface surveys.

[0028] Each observation point is equipped with an intelligent data acquisition station responsible for controlling sensors, performing analog-to-digital conversion, data packaging, local caching, and preliminary processing (such as digital filtering and event detection triggering). It features high-precision clock synchronization (e.g., PTP or GPS timekeeping: using a GPS-disciplined, high-stability, temperature-controlled crystal oscillator OCXO as the time reference for each node, and achieving microsecond-level intra-network synchronization via the Network Time Protocol (PTP), local storage, preliminary data processing (such as downsampling and filtering), and event triggering capabilities. Compressed continuous data or triggered event data is transmitted in real-time to the central server via a dedicated fiber optic network, wireless mesh network, or satellite communication. Specifically, a layered hybrid network architecture can be adopted. Downhole stations aggregate data to the station gateway via a wireless self-organizing network or fiber optic cable; the gateway transmits data in real-time to the remote data center via a backbone dedicated fiber optic network or high-bandwidth wireless links (such as microwave or satellite). It supports breakpoint resume and bandwidth adaptive transmission. Specifically, an integrated three-dimensional observation architecture is constructed based on air-assisted monitoring units, ground monitoring units, and downhole monitoring units. This architecture enables the coordinated acquisition of surface deformation, near-field high signal-to-noise ratio microseismic signals, and deep formation parameters. It compensates for the shortcomings of strong signal attenuation of a single ground station and the limited coverage of a single downhole array. This allows for blind-spot-free monitoring of key risk areas in the target gas storage area, providing a high-quality data source for subsequent high-precision positioning and mechanical inversion.

[0029] As one possible approach, the well-to-surface joint observation optimization scheme involved in step S110 is obtained through the following steps: The three-dimensional geomechanical model of the target gas storage facility, the lower limit of the target monitoring magnitude, the preset risk area, and the well-ground engineering constraints are obtained. The well-ground engineering constraints include the upper limit of the number of surface stations and monitoring wells, the upper limit of the number of downhole sensors, the wellhead location limit, the cable length limit, and the deployment cost constraint. Based on a three-dimensional geomechanical model, the lower limit of target monitoring magnitude, and a preset risk area, a multi-objective optimization function is constructed with positioning accuracy and spatial coverage as the core. The optimization objective of positioning accuracy is to maximize the determinant value of the Fisher information matrix, and the optimization objective of spatial coverage is to maximize the theoretically detectable spatial coverage. A multi-objective optimization problem is constructed based on a multi-objective optimization function and well-ground engineering constraints. A non-dominated sorting genetic algorithm with an elitist strategy is used to iteratively solve the multi-objective optimization problem, and the Pareto optimal solution set is obtained (i.e., all deployment schemes that cannot improve one objective without excessively harming another objective). Based on actual engineering needs, the optimal implementation scheme is selected from the Pareto optimal solution set to obtain the well-ground joint observation optimization scheme; The optimized well-ground joint observation scheme includes at least the latitude and longitude coordinates and layout of the ground stations, the location coordinates and number of monitoring wells, and the depth range and level spacing of the downhole sensor strings; the layout can be either a regular grid or a surrounding cavity.

[0030] The high-precision three-dimensional geomechanical model includes stratigraphic structure, velocity model, salt cavity geometry, known faults, etc.; the preset risk areas specifically include the cavity roof, fault zones, weak caprock zones, etc.

[0031] Objective Function 1 (Accuracy Objective): Maximize the determinant of the Fisher information matrix F (D-optimality criterion). The F matrix is ​​determined by the geometric and velocity model of the ray paths from the event to each observation point, and the trace of its inverse matrix is ​​related to the volume of the positioning error ellipsoid. Maximizing det(F) means statistically minimizing the average positioning error.

[0032] Objective Function 2 (Coverage Objective): Maximize the theoretically detectable spatial coverage. Based on the acoustic radiation model and noise level, calculate the detectable magnitude of each potential observation point relative to the spatial grid points, and statistically analyze the proportion of grid points with detectable magnitudes lower than the target magnitude.

[0033] Step S120: Perform microseismic event detection and three-dimensional joint localization on the physical field monitoring data to obtain the microseismic event localization results; As one possible approach, physical field monitoring data includes continuous waveform data. The specific process of detecting microseismic events from physical field monitoring data involved in step S120 above includes: Continuous waveform data is entered into a message queue and sliced ​​into waveform slices at fixed intervals to obtain waveform slice data. The waveform slice data is input into the trained improved U-Net convolutional neural network model to obtain the probability values ​​of P-wave, S-wave and noise output point by point along the time axis, which constitute a time-series probability sequence. Suspected microseismic event segments are extracted based on the time-series probability sequence. The suspected microseismic event fragments and their corresponding original continuous waveform data are input into the Transformer encoder. The authenticity of the event is determined by the self-attention mechanism to obtain the real microseismic events. The first arrival times of the P-wave and S-wave are picked up for the real microseismic events to obtain the first arrival times of the P / S-wave for each real microseismic event.

[0034] Specifically, such as Figure 4As shown, an improved U-Net convolutional neural network is first employed, using continuous waveform data segments with single or limited channels as input images (time series as height, channels as width / color channels). The network outputs a probability sequence for each time point belonging to a P-wave, S-wave, or noise. This network excels at capturing the local spatiotemporal features of signals, quickly filtering out a large number of segments containing potential events. Then, all suspected segments output from the previous stage, along with their corresponding original channel data, are input into a Transformer-based encoder network. The Transformer's self-attention mechanism effectively models long-distance dependencies between multiple channels, comprehensively judging whether a segment is a real event, and precisely picking the first arrival times of P-waves and S-waves with subsampling accuracy. This two-stage structure significantly reduces the false alarm rate while maintaining high recall.

[0035] In other words, this application uses a two-level cascaded AI model to achieve highly reliable detection of low signal-to-noise ratio micro-seismic events, overcoming the problems of missed detection and false detection caused by gas storage injection and production noise and environmental interference. It can achieve accurate picking of P-waves and S-waves at the sub-sampling level, significantly reducing the false alarm rate in actual data processing, providing accurate time-to-time data for subsequent high-precision three-dimensional positioning, and supporting automated and real-time monitoring.

[0036] As one possible approach, such as Figure 5 As shown, the specific process of three-dimensional joint positioning involved in step S120 above includes: Based on the first arrival times of P-waves and S-waves of each real microseismic event, the simplex method is used to locate each real microseismic event and obtain preliminary location results. Based on the waveform cross-correlation coefficients between each real microseismic event, the real microseismic events are correlated and paired to obtain multiple event pairs; Construct double-difference localization equations for all event pairs and solve them to obtain the location correction parameters for each real microseismic event; The initial location results of the actual microseismic events are corrected by using location correction parameters to obtain the location results of the microseismic events.

[0037] As one possible approach, the double-difference localization equation for each pair of events is: ; in, For events in an event pair and events Reach any observation point The travel time difference of P / S waves, where the observation point is any one of the ground station and the monitoring well; For the event Arrive at the observation point The travel time of the P / S wave; For the event Arrive at the observation point P / S wave travel time; It is an event and events Spatial location difference, It is an event and events The time difference of the earthquake's occurrence; For the event The travel-time partial derivative; For the event The traveling partial derivatives.

[0038] Specifically, by eliminating the systematic deviation caused by the velocity model and ray path through time difference, the positioning equation is made more stable and reliable, accurately reflecting the spatial relative position between microseismic events, and significantly improving the relative positioning accuracy, which can reach the meter level.

[0039] Step S130: Based on the microseismic event location results and physical field monitoring data, perform source mechanism inversion and regional stress field multi-scale inversion to obtain mechanical parameter inversion results; As one possible approach, the specific process of source mechanism inversion involved in step S130 above includes: Based on the microseismic event location results, valid microseismic events with a sufficient number of effective records at observation points and complete waveform data were selected. Based on the location results of the effective microseismic events, six sets of basic Green's functions corresponding to each observation point of the effective microseismic event are obtained from the pre-constructed three-dimensional Green's function database. The six sets of basic Green's functions correspond one-to-one with the six independent components of the moment tensor. For each effective microseismic event, the theoretical simulated waveform is obtained by convolving each set of basic Green's functions with the corresponding moment tensor components. With the goal of minimizing the L2 norm residual between the measured waveform and the theoretical simulated waveform, the optimal moment tensor for each effective microseismic event is obtained by iterative optimization using the adjoint state method combined with the L-BFGS algorithm. The optimal moment tensor is orthogonally decomposed to obtain the source rupture characteristic parameters of each effective microseismic event; The three-dimensional Green's function database is constructed according to the following steps: the study area of ​​the three-dimensional geomechanical model is divided into grids according to space and depth to obtain multiple grid nodes; based on the three-dimensional velocity model, the spectral element method is used to calculate the six sets of basic Green's functions corresponding to each observation point of each grid node; and the Green's functions of all grid nodes are collected to form a three-dimensional Green's function database covering the study area.

[0040] The source of a microearthquake can be described by a second-order symmetric moment tensor M. The displacement waveform observed at station n. It can be represented as a Green's function With moment tensor Convolution: Green's function describes the wave propagation path effect from the source to the station.

[0041] An inversion framework based on the adjoint state method is adopted. First, using the accurate velocity model and positioning results, six sets of fundamental Green's functions (corresponding to the six independent components of the moment tensor) are calculated for each event-station pair. The inversion objective is to minimize the L2 norm residuals between all station observed waveforms and theoretical waveforms. By calculating the gradient of the objective function with respect to the moment tensor components, optimization algorithms such as L-BFGS are used to efficiently solve for the optimal moment tensor M.

[0042] The solved moment tensor can be decomposed into isotropic (ISO), compensated linear vector dipole (CLVD), and dual couple (DC) components. The higher the proportion of the dual couple component, the closer it is to pure shear fracture; a significant isotropic component may indicate tensile or compressive fracturing. From the dual couple components, the strike, dip, and slip angle of the fault can be further obtained.

[0043] This application employs full-waveform moment tensor inversion to achieve high-precision solution of source mechanical parameters. The one-to-one correspondence between the six sets of Green's functions and moment tensor components ensures the accuracy of theoretical waveform fitting. The adjoint state method and L-BFGS significantly improve inversion efficiency and convergence speed.

[0044] As one possible approach, the specific process of multi-scale inversion of the regional stress field involved in step S130 above includes: Based on the microseismic event location results, the effective microseismic events are divided into three time series subsets: the pre-pressurization period, the rapid pressurization period, and the pressure stabilization period. Based on the Wallace-Bott mechanical assumptions, and constrained by the source rupture characteristic parameters of microseismic events within each time series subset, the Bayesian MCMC method is used to invert the stress tensor of the corresponding region in the corresponding time period, and the stress field inversion results of each time series stage are obtained. The mechanical parameter inversion results include the stress field inversion results and the source rupture characteristic parameters.

[0045] Assumption: Within a sufficiently small time and space window, the background stress field that causes a set of microseismic events is uniform.

[0046] Model: Based on the Wallace-Bott assumption, the fault slip direction is parallel to the shear stress direction on the fault plane. A uniform stress tensor is given. This allows us to predict the optimal slip direction on each fault plane (obtained through moment tensor inversion). Inversion involves finding a... This minimizes the objective function between the predicted sliding direction and the actual inverted sliding direction.

[0047] The Bayesian MCMC method is used to invert the stress tensor of the corresponding region for the corresponding time period. Specifically, the Markov Chain Monte Carlo (MCMC) method is used to analyze the stress tensor parameters (three principal stress directions). , , and stress shape factor Sampling is performed. This method not only provides an optimal estimate of stress but also offers a complete posterior probability distribution, quantifying the uncertainty of the parameters. Through sliding time window analysis, the evolution of the stress field direction and tensor ratio over time can be tracked.

[0048] Step S140: Based on the microseismic event location results, mechanical parameter inversion results, and physical field monitoring data, conduct dynamic risk assessment and graded early warning.

[0049] As one possible approach, the physical field monitoring data also includes engineering monitoring data. The dynamic risk assessment and graded early warning based on the microseismic event location results, the mechanical parameter inversion results, and the physical field monitoring data involved in step S140 above specifically includes the following: Multi-dimensional indicators were extracted from microseismic event location results, mechanical parameter inversion results, and physical field monitoring data. Specifically, these included: event frequency, moment release rate, b-value, spatial clustering of events, and event depth migration trend from microseismic event location results; proportion of double couples, proportion of isotropic components, and consistency of P-axis direction from source rupture characteristic parameters; rotation angle of the maximum principal stress direction, stress shape factor change, and ratio of shear stress to normal stress from stress field inversion results; and storage pressure, injection-production rate, pressure change rate, and cumulative injection-production volume of the target gas storage facility from engineering monitoring data. After normalizing the multi-dimensional indicators, they are input into a preset dynamic risk assessment model to calculate the comprehensive risk index. The comprehensive risk index is compared with the preset three-level thresholds of attention, early warning and alarm. When the comprehensive risk index exceeds the corresponding threshold or a single indicator shows extreme abnormality, the corresponding level of early warning is triggered. It outputs early warning information, risk locations, risk mechanism analysis reports, and 3D visualization scenes to provide decision support for the operation and control of target gas storage facilities.

[0050] Specifically, quantitative indicators are extracted from the following four dimensions: Extracting event frequency from microseismic event location results Seismic moment release rate , Values ​​(proportion of earthquakes of different sizes), spatial clustering of events (e.g., density-based DBSCAN clustering analysis), and event depth migration trends; Extract the proportion of dual couple (DC), the proportion of isotropic (ISO) components (positive values ​​indicate tensile cracking), the consistency of the P-axis (compression axis) direction (reflecting the stability of the regional stress state), and the spatiotemporal evolution of the source mechanism type from the characteristic parameters of the source rupture. Extract the maximum principal stress from the stress field inversion results. Orientation rotation angle, stress shape factor The changes in shear stress and the ratio of inverted shear stress to normal stress; Extract reservoir pressure P, injection-production rate Q, and pressure change rate from engineering monitoring data. Cumulative gas injection and production volume.

[0051] Construct a comprehensive risk index The dynamic risk assessment model can take the form of a weighted linear combination or a more complex machine learning model (such as random forests or neural networks). For example: ; in, , , , For each index, the normalization or nonlinear transformation function, the weights The model is obtained through analysis of historical anomalies, numerical simulation, or expert knowledge calibration. It can be updated online.

[0052] By setting multi-level early warning thresholds (e.g., attention, warning, alarm). When... Exceeding the threshold, or a single indicator showing extreme abnormalities (e.g.) When a value drops sharply or a large number of tensile events occur, the system automatically triggers an alarm of the corresponding level. The warning information, along with a detailed data analysis report and a 3D visualization scene, is pushed to the decision-maker's terminal, providing the possible location and mechanism analysis of the risk source.

[0053] The 3D visualization application scenarios are as follows: Web-based 3D visualization engine: built using Cesium.js, Three.js, or Unity / Unreal Engine. It can load 3D geological models, salt cavity models, and fault models, and dynamically render: micro-seismic event points playing over time (color represents magnitude, shape represents mechanism); display of the "beach ball" symbol of the focal mechanism in 3D space; spatial distribution and evolution of the stress tensor ellipsoid; overlay of risk heat maps on cavity surfaces or stratigraphic interfaces; real-time engineering data curves (pressure, flow) linked to seismic activity parameters in a linked chart; early warning information dashboard: displaying the comprehensive risk index, status of each sub-indicator, current warning level, latest alarm event list, and handling suggestions in the form of a large-screen dashboard. It supports receiving critical alarms and viewing brief reports and 3D scenes on tablets and mobile phones.

[0054] Compared with the prior art, the present invention has the following significant advantages: (1) A qualitative leap in monitoring performance: Through the optimized design of the observation system, the sensitivity and positioning accuracy of the monitoring network have been improved from the source, ensuring "no blind spots" monitoring of key risk areas and reducing the lower limit of detectable magnitude by at least 0.5 magnitude units.

[0055] (2) Highly automated and intelligent data processing: Based on deep learning, event detection and picking overcome the limitations of traditional methods under high noise, realize unattended automated processing, minimize human intervention, and improve processing efficiency by more than an order of magnitude.

[0056] (3) The positioning accuracy meets the practical requirements of engineering: the dual-difference positioning method with integrated waveform cross-correlation effectively overcomes the systematic error caused by the inaccuracy of the velocity model, stabilizes the relative positioning accuracy of microseismic events within 10 meters, and the absolute positioning accuracy is better than 30 meters, which is sufficient to accurately characterize the crack zone and determine the distance between the rupture and the sealing layer.

[0057] (4) It has achieved a leap from "monitoring" to "diagnosis": full waveform moment tensor inversion and stress field inversion provide information on rupture mechanisms and stress states that cannot be obtained by traditional monitoring. It can clearly distinguish between shear rupture (which may be related to fault activity) and tensile rupture (which may be related to leakage risk), providing a solid physical and mechanical basis for risk assessment.

[0058] (5) Scientific and quantitative risk assessment: The constructed multi-parameter fusion dynamic risk model has changed the previous qualitative judgment mode that relied on a single indicator or experience, and realized the quantitative, comprehensive and dynamic evaluation of the integrity status of the gas storage facility, which greatly improves the accuracy and foresight of the early warning.

[0059] (6) Advanced system architecture and strong scalability: The system adopts a cloud-edge-device collaborative architecture and microservice design, which makes the system elastic, easy to maintain and upgrade. New algorithms and new sensors can be easily integrated into the existing framework.

[0060] (7) Support for full life cycle management: The method and system of the present invention can be adjusted and emphasized according to the characteristics of different stages of gas storage construction, operation and disposal, forming a complete management technology system that runs through the entire process and has long-term application value.

[0061] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below in conjunction with the deployment and application of a safety monitoring system for a salt cavern gas storage facility, such as... Figure 6-7 As shown. The following detailed implementation process is only used to explain the present invention and is not intended to limit the present invention.

[0062] 1. Project Background: A salt cavern gas storage facility has a designed capacity of 500 million cubic meters, a maximum operating pressure of 20 MPa, and a minimum operating pressure of 8 MPa. The cavity is approximately 1100 meters deep and is covered by a 300-meter-thick mudstone caprock. A known small fault exists nearby, approximately 500 meters from the nearest cavity. The monitoring objectives are: real-time monitoring of cavity creep and potential rupture, assessment of caprock integrity, monitoring for signs of fault activation, and early warning of risks that may arise from excessive pressure changes.

[0063] 2. Implementation Steps Step S1: Observation network optimization design: Input preparation: Establish a detailed three-dimensional geological velocity model: Integrate well logging, VSP, and surface seismic data to construct a velocity model that includes salt layers, caprock, and fault fracture zones, and consider the velocity differences and anisotropy between salt rock and surrounding rock for preliminary estimation.

[0064] Define the target monitoring area: with a boundary of 200 meters outward from all salt cavities, extending upward to the ground surface and downward to 100 meters below the bottom of the salt cavities.

[0065] Target performance is set as follows: the probability of detecting microseismic events with ML≥-1.0 within the target area is >90%, and the location error (RMS) is <20 meters.

[0066] Engineering constraints: A maximum of two dedicated monitoring wells (reaching the top of the salt layer) can be drilled; no more than 30 monitoring stations can be deployed on the surface; the total budget constraint is... .

[0067] Optimization calculation: The possible locations of the surface area and two potential monitoring wells are discretized into candidate points. The NSGA-II optimization algorithm library is invoked. For each individual in each generation of the population (representing a combination of station and downhole sensor depth scheme): 1000 test source points are randomly generated within the target area. The travel time from each test point to all stations is calculated using ray tracing, and the Fisher information matrix F is constructed and accumulated, and det(F) is calculated. Based on the preset noise level and radiation pattern, the detection capability of each station for each grid point is calculated, and the coverage is statistically analyzed. The total cost of the scheme is calculated. After thousands of iterations, the Pareto optimal front is obtained.

[0068] Solution selection: Choose one from the Pareto fronts that is within budget and has good positioning accuracy. A high-resolution solution with over 95% coverage. The final design is as follows: Figure 6 As shown, specifically: 20 ground stations are deployed (distributed in a double ring around the cavity group), and 2 monitoring wells are drilled on the east and west sides of the cavity group. Each well is equipped with a 16-level three-component geophone string, with a depth range of 800 meters to 1150 meters (from the bottom of the caprock to the top of the salt cavity).

[0069] Step S2: System construction and data collection, such as Figure 7 As shown: Twenty ground stations were constructed according to the design coordinates, and equipment was installed and networked. Two monitoring wells were drilled, and after completion, geophone strings and DAS optical fibers were installed at the wellheads. The time synchronization system of all nodes was calibrated to ensure a synchronization error of <1 ms. Continuous data acquisition began, with a sampling rate of 500 Hz for the ground stations and 2000 Hz for the downhole geophones. Data was transmitted in real time to a data center 50 kilometers away from the site via a fiber optic ring network.

[0070] Step S3: Intelligent detection and positioning (taking a single gas injection and pressurization process as an example) Data flow: Continuous waveform data enters the Kafka message queue.

[0071] The intelligent detection service was triggered: U-Net detection: A 1-minute data slice per channel was fed into the trained U-Net model. The model output identified 5 segments containing suspected P-wave vibrations.

[0072] Transformer precision analysis: The raw data (52 channels in total) from all 20 + 32 channels corresponding to these 5 segments are sliced ​​and fed into the Transformer model. The model determines that 3 of them are real microseismic events and outputs accurate P-wave and S-wave arrival times (accuracy of approximately 0.001 seconds).

[0073] Location service triggered: Initial absolute location values: Using the simplex method and the time difference between P-wave and S-wave arrival, the preliminary locations of the three events are given, all located in the surrounding rock on the northwest side of cavity A, at a depth of approximately 1050 meters.

[0074] Double-difference localization: The waveform cross-correlation between these three events and between them and 20 other events occurring in the same region over the past week was calculated to construct event pairs with high correlation (>0.8). In the 3D anisotropic velocity model, the travel-time partial derivatives of all event-station pairs were calculated using Fast Mover Method (FMM) ray tracing. The double-difference linear equation system was constructed and solved.

[0075] Results: Fine-grained point clouds of 23 events were obtained. The results show that the events are not randomly distributed, but rather exhibit two clear linear clusters: a nearly vertical cluster located on the side wall of cavity A, and a nearly horizontal cluster located at the bottom of the cap layer. The average semi-major axis of the positioning error ellipse is 8 meters.

[0076] Step S4: Focal Mechanism and Stress Inversion Moment tensor inversion: Of the 23 located events, those clearly recorded by at least 15 stations were selected for inversion (18 in total). A three-dimensional Green's function database for the study area (for grid points of different depths) was pre-calculated using the spectral element method (SEM). For each event, full waveform data of its P-wave and S-wave windows were extracted, and its moment tensor was inverted using the adjoint state method.

[0077] Results analysis: Near-vertical cluster events located on the sidewalls of the cavity exhibit a moment tensor dominated by double couples (DC), accounting for >80%, indicating shear slip and possibly related to local stress adjustments caused by salt creep. Near-horizontal cluster events located at the bottom of the caprock show a significant positive isotropic (ISO) component (20%-40%) mixed with double couples in their moment tensor, indicating tensile fracturing components, possibly related to increased gas injection pressure leading to tension at the bottom of the caprock.

[0078] Stress field inversion: The 23 events were divided into three groups according to time (pre-pressurization period, rapid pressurization period, and pressure stabilization period). For each group of events, the stress tensor of the region during that period was inverted using the Bayesian MCMC method, utilizing the focal mechanism solution (dual couple part).

[0079] Results analysis: It was found that from the early stage of pressurization to the rapid pressurization period, the maximum principal stress... The direction of the stress rotated clockwise by about 15 degrees, and the stress shape factor R increased slightly, indicating that the stress state changed from a relatively stable triaxial state to a state more inclined towards uniaxial compression, which is consistent with the increase in horizontal stress caused by injection.

[0080] Step S5: Risk Assessment and Early Warning Indicator calculation: The event rate N increased fivefold during the rapid increase in pressure. The value decreased from 1.2 to 0.8. The average ISO% of the event group at the bottom of the caprock was 25%. Rotate 15 degrees in direction. Rate of change of pressure in the reservoir cavity. The standard pressure is +2 MPa / day. Comprehensive risk index calculation: After normalizing the above indicators, input them into the trained random forest risk assessment model. The model outputs the current comprehensive risk index. (Range 0-1).

[0081] Early warning decision: System preset thresholds: Attention level 0.4, Early warning level 0.6, Alarm level 0.8. Currently... This triggers a warning level (orange) alert.

[0082] Early warning output: The decision support terminal's large screen flashes an orange alert, and a pop-up window displays: "Increased tensile fracturing activity detected at the bottom of the caprock, accompanied by stress field disturbance. Close monitoring and control of the gas injection rate are recommended." The 3D visualization scene automatically focuses on the event cluster at the bottom of the caprock and highlights its tensile mechanism symbol. The system automatically generates a report, including an event list, location map, focal mechanism solution, stress inversion results, and detailed risk analysis, and sends it to the mobile application of the gas storage facility's operations manager and engineers.

[0083] Follow-up actions: Based on the early warning, operators reduced the gas injection rate by 30% from the original plan and strengthened monitoring of pressure and related indicators. In the following days, microseismic activity subsided, and the risk index fell below the concern level, verifying the effectiveness of the early warning.

[0084] 3. System Architecture Implementation Cloud platform deployment: All microservices are deployed using a Kubernetes cluster in the data center. The data lake is built on HDFS and object storage.

[0085] Visual development: Cesium.js is used to build a web-based 3D application, which communicates with the backend via WebSocket and REST API to achieve dynamic data loading.

[0086] Security measures: The system network is isolated by firewalls and access is controlled via VPN; data is encrypted throughout transmission and storage; and strict user access management is implemented.

[0087] The implementation of this embodiment demonstrates that the present invention can effectively monitor microseismic activity under complex operating conditions of gas storage facilities and successfully transform the monitoring data into risk warnings with clear engineering guidance significance, greatly enhancing the initiative and scientific nature of gas storage facility safety management.

[0088] The following describes an embodiment of the gas storage safety monitoring device based on well-ground joint observation optimization according to the present invention. Please refer to [link / reference]. Figure 8 , Figure 8 This is a schematic diagram of an embodiment of a gas storage safety monitoring device optimized based on well-ground joint observation in this invention. The device 800 includes: The signal acquisition unit 801 is used to acquire physical field monitoring data of the target gas storage facility based on the three-dimensional observation network constructed at the target gas storage facility. The detection and positioning unit 802 is used to perform microseismic event detection and three-dimensional joint positioning on physical field monitoring data to obtain microseismic event positioning results. Data inversion unit 803 is used to perform source mechanism inversion and regional stress field multi-scale inversion based on microseismic event location results and physical field monitoring data to obtain mechanical parameter inversion results; The assessment and early warning unit 804 is used for dynamic risk assessment and graded early warning based on microseismic event location results, mechanical parameter inversion results and physical field monitoring data.

[0089] As an example of implementation, the three-dimensional observation network consists of an airborne auxiliary monitoring unit, a ground monitoring unit, and a downhole monitoring unit. The airborne auxiliary monitoring unit uses satellite InSAR and UAV survey equipment to conduct periodic surface monitoring. The ground monitoring unit includes several ground stations, and the downhole monitoring unit includes several monitoring wells and downhole sensor strings. The ground stations, monitoring wells, and downhole sensor strings are deployed according to the well-ground joint observation optimization scheme.

[0090] As an exemplary implementation, the signal acquisition unit 801 is further configured to perform the following steps: The three-dimensional geomechanical model of the target gas storage facility, the lower limit of the target monitoring magnitude, the preset risk area, and the well-ground engineering constraints are obtained. The well-ground engineering constraints include the upper limit of the number of surface stations and monitoring wells, the upper limit of the number of downhole sensors, the wellhead location limit, the cable length limit, and the deployment cost constraint. Based on a three-dimensional geomechanical model, the lower limit of target monitoring magnitude, and a preset risk area, a multi-objective optimization function is constructed with positioning accuracy and spatial coverage as the core. The optimization objective of positioning accuracy is to maximize the determinant value of the Fisher information matrix, and the optimization objective of spatial coverage is to maximize the theoretically detectable spatial coverage. A multi-objective optimization problem is constructed based on a multi-objective optimization function and well-ground engineering constraints. A non-dominated sorting genetic algorithm with an elitist strategy is used to iteratively solve the multi-objective optimization problem and obtain a Pareto optimal solution set. Based on actual engineering needs, the optimal implementation scheme is selected from the Pareto optimal solution set to obtain the well-ground joint observation optimization scheme; The optimized well-ground joint observation scheme includes at least the latitude and longitude coordinates and layout of the ground stations, the location coordinates and number of monitoring wells, and the depth range and level spacing of the downhole sensor strings; the layout can be either a regular grid or a surrounding cavity.

[0091] As an example implementation, the physical field monitoring data includes continuous waveform data, and the detection and positioning unit 802 is further configured to perform the following steps: Continuous waveform data is entered into a message queue and sliced ​​into waveform slices at fixed intervals to obtain waveform slice data. The waveform slice data is input into the trained improved U-Net convolutional neural network model to obtain the probability values ​​of P-wave, S-wave and noise output point by point along the time axis, which constitute a time-series probability sequence. Suspected microseismic event segments are extracted based on the time-series probability sequence. The suspected microseismic event fragments and their corresponding original continuous waveform data are input into the Transformer encoder. The authenticity of the event is determined by the self-attention mechanism to obtain the real microseismic events. The first arrival times of the P-wave and S-wave are picked up for the real microseismic events to obtain the first arrival times of the P / S-wave for each real microseismic event.

[0092] As an exemplary implementation, the detection and positioning unit 802 is further configured to perform the following steps: Based on the first arrival times of P-waves and S-waves of each real microseismic event, the simplex method is used to locate each real microseismic event and obtain preliminary location results. Based on the waveform cross-correlation coefficients between each real microseismic event, the real microseismic events are correlated and paired to obtain multiple event pairs; Construct double-difference localization equations for all event pairs and solve them to obtain the location correction parameters for each real microseismic event; The initial location results of the actual microseismic events are corrected by using location correction parameters to obtain the location results of the microseismic events.

[0093] As an example implementation, the double-difference localization equation for each event pair is: ; in, For events in an event pair and events Reach any observation point The travel time difference of P / S waves, where the observation point is any one of the ground station and the monitoring well; For the event Arrive at the observation point The travel time of the P / S wave; For the event Arrive at the observation point P / S wave travel time; It is an event and events Spatial location difference, It is an event and events The time difference of the earthquake's occurrence; For the event The travel-time partial derivative; For the event The traveling partial derivatives.

[0094] As an exemplary implementation, the data inversion unit 803 is further configured to perform the following steps: Based on the microseismic event location results, valid microseismic events with a sufficient number of effective records at observation points and complete waveform data were selected. Based on the location results of the effective microseismic events, six sets of basic Green's functions corresponding to each observation point of the effective microseismic event are obtained from the pre-constructed three-dimensional Green's function database. The six sets of basic Green's functions correspond one-to-one with the six independent components of the moment tensor. For each effective microseismic event, the theoretical simulated waveform is obtained by convolving each set of basic Green's functions with the corresponding moment tensor components. With the goal of minimizing the L2 norm residual between the measured waveform and the theoretical simulated waveform, the optimal moment tensor for each effective microseismic event is obtained by iterative optimization using the adjoint state method combined with the L-BFGS algorithm. The optimal moment tensor is orthogonally decomposed to obtain the source rupture characteristic parameters of each effective microseismic event; The three-dimensional Green's function database is constructed according to the following steps: the study area of ​​the three-dimensional geomechanical model is divided into grids according to space and depth to obtain multiple grid nodes; based on the three-dimensional velocity model, the spectral element method is used to calculate the six sets of basic Green's functions corresponding to each observation point of each grid node; and the Green's functions of all grid nodes are collected to form a three-dimensional Green's function database covering the study area.

[0095] As an exemplary implementation, the data inversion unit 803 is further configured to perform the following steps: Based on the microseismic event location results, the effective microseismic events are divided into three time series subsets: the pre-pressurization period, the rapid pressurization period, and the pressure stabilization period. Based on the Wallace-Bott mechanical assumptions, and constrained by the source rupture characteristic parameters of microseismic events within each time series subset, the Bayesian MCMC method is used to invert the stress tensor of the corresponding region in the corresponding time period, and the stress field inversion results of each time series stage are obtained. The mechanical parameter inversion results include the stress field inversion results and the source rupture characteristic parameters.

[0096] As an exemplary implementation, the evaluation and early warning unit 804 is further configured to perform the following steps: The physical field monitoring data also includes engineering monitoring data. Based on the microseismic event location results, the mechanical parameter inversion results, and the physical field monitoring data, dynamic risk assessment and graded early warning are performed, including: Multi-dimensional indicators were extracted from microseismic event location results, mechanical parameter inversion results, and physical field monitoring data. Specifically, these included: event frequency, moment release rate, b-value, spatial clustering of events, and event depth migration trend from microseismic event location results; proportion of double couples, proportion of isotropic components, and consistency of P-axis direction from source rupture characteristic parameters; rotation angle of the maximum principal stress direction, stress shape factor change, and ratio of shear stress to normal stress from stress field inversion results; and storage pressure, injection-production rate, pressure change rate, and cumulative injection-production volume of the target gas storage facility from engineering monitoring data. After normalizing the multi-dimensional indicators, they are input into a preset dynamic risk assessment model to calculate the comprehensive risk index. The comprehensive risk index is compared with the preset three-level thresholds of attention, early warning and alarm. When the comprehensive risk index exceeds the corresponding threshold or a single indicator shows extreme abnormality, the corresponding level of early warning is triggered. It outputs early warning information, risk locations, risk mechanism analysis reports, and 3D visualization scenes to provide decision support for the operation and control of target gas storage facilities.

[0097] In this embodiment of the invention, firstly, physical field monitoring data of the target gas storage facility is collected by the signal acquisition unit 801 through a three-dimensional observation network constructed at the target gas storage facility; then, microseismic event detection and three-dimensional joint positioning are performed on the physical field monitoring data by the detection and positioning unit 802 to obtain the microseismic event positioning results; secondly, the source mechanism inversion and regional stress field multi-scale inversion are carried out by the data inversion unit 803 based on the microseismic event positioning results and physical field monitoring data to obtain the mechanical parameter inversion results; finally, the dynamic risk assessment and graded early warning are carried out by the assessment and early warning unit 804 based on the microseismic event positioning results, mechanical parameter inversion results, and physical field monitoring data. This application improves monitoring sensitivity and positioning accuracy from the source by constructing a three-dimensional observation network; it achieves unattended automated processing through event detection and picking, minimizing human intervention; it provides rupture mechanism and stress state information that traditional monitoring cannot obtain through full-waveform moment tensor inversion and stress field inversion, providing a solid physical and mechanical foundation for risk assessment; through the complete process of three-dimensional observation network acquisition, intelligent detection and positioning, multi-scale mechanical inversion and multi-parameter risk early warning, it realizes intelligent monitoring of the entire chain of gas storage facilities from data perception to mechanical diagnosis to decision-making and early warning, achieving the safety monitoring requirements of high precision, high stability and full-process automation for gas storage facilities, as well as effective coverage of key risk areas and accurate diagnosis of rupture mechanisms.

[0098] This invention also provides an electronic device, please refer to [link / reference]. Figure 9 , Figure 9 This is a schematic diagram of one embodiment of the electronic device of the present invention, including: The system includes a memory 901, a processor 902, and a computer program 903 stored in the memory and executable on the processor. When the processor executes the computer program 903 stored in the memory, it implements the above-mentioned gas storage safety monitoring method based on well-ground joint observation optimization.

[0099] For ease of explanation, only the parts relevant to the embodiments of the present invention are shown. For specific technical details not disclosed, please refer to the part of the gas storage safety monitoring method based on well-ground joint observation optimization in the embodiments of the present invention. The memory 901 can be used to store the computer program 903, which includes software programs, modules, and data. The processor 902 executes the computer program 903 stored in the memory 901 to perform various functional applications and data processing of the electronic device.

[0100] This invention also provides a computer-readable storage medium; please refer to [link to relevant documentation]. Figure 10 , Figure 10This is a schematic diagram of an embodiment of a computer-readable storage medium in the present invention. The computer-readable storage medium may store a computer program, which, when executed, includes some or all of the steps of the gas storage safety monitoring method based on well-ground joint observation optimization described in the above method embodiments.

[0101] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, electronic equipment, and computer-readable storage media described above can be referred to the corresponding processes of the gas storage safety monitoring method based on well-ground joint observation optimization in the foregoing method embodiments, and will not be repeated here.

[0102] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0103] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0104] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0105] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the gas storage safety monitoring method based on well-ground joint observation optimization according to various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0106] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A gas storage safety monitoring method based on well-ground joint observation optimization, characterized in that, The method includes: Physical field monitoring data of the target gas storage facility are collected based on a three-dimensional observation network constructed at the target gas storage facility. Microseismic event detection and three-dimensional joint localization are performed on the physical field monitoring data to obtain microseismic event localization results; Based on the microseismic event location results and the physical field monitoring data, source mechanism inversion and regional stress field multi-scale inversion were carried out to obtain mechanical parameter inversion results. Dynamic risk assessment and graded early warning are conducted based on the microseismic event location results, the mechanical parameter inversion results, and the physical field monitoring data.

2. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 1, characterized in that, The three-dimensional observation network consists of an air-assisted monitoring unit, a ground monitoring unit, and a downhole monitoring unit. The air-assisted monitoring unit uses satellite InSAR and UAV survey equipment to conduct periodic surface monitoring. The ground monitoring unit includes several ground stations, and the downhole monitoring unit includes several monitoring wells and downhole sensor strings. The ground stations, monitoring wells, and downhole sensor strings are deployed according to the well-ground joint observation optimization scheme.

3. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 2, characterized in that, The optimized well-ground joint observation scheme is obtained through the following steps: The three-dimensional geomechanical model of the target gas storage facility, the lower limit of the target monitoring magnitude, the preset risk area, and the well-ground engineering constraints are obtained. The well-ground engineering constraints include the upper limit of the number of surface stations and monitoring wells, the upper limit of the number of downhole sensors, the wellhead location limit, the cable length limit, and the deployment cost constraint. Based on the aforementioned three-dimensional geomechanical model, the lower limit of the target monitoring magnitude, and the preset risk area, a multi-objective optimization function is constructed with positioning accuracy and spatial coverage as the core. The optimization objective of positioning accuracy is to maximize the determinant value of the Fisher information matrix, and the optimization objective of spatial coverage is to maximize the theoretically detectable spatial coverage. A multi-objective optimization problem is constructed based on the multi-objective optimization function and well-ground engineering constraints. The multi-objective optimization problem is solved iteratively using a non-dominated sorting genetic algorithm with an elite strategy to obtain a Pareto optimal solution set. Based on actual engineering requirements, the optimal implementation scheme is selected from the Pareto optimal solution set to obtain the well-ground joint observation optimization scheme; The well-ground joint observation optimization scheme includes at least the latitude and longitude coordinates and layout of the ground stations, the location coordinates and number of monitoring wells, and the depth range and level spacing of the downhole sensor strings; the layout can be either a regular grid or a surrounding cavity.

4. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 1, characterized in that, The physical field monitoring data includes continuous waveform data, and the specific process of detecting microseismic events using the physical field monitoring data includes: Continuous waveform data is entered into a message queue and sliced ​​according to a fixed duration to obtain waveform slice data. The waveform slice data is input into the trained improved U-Net convolutional neural network model to obtain the probability values ​​of P-wave, S-wave and noise output point by point along the time axis, which constitute a time-series probability sequence. Suspected microseismic event segments are extracted based on the time-series probability sequence. The suspected microseismic event fragment and its corresponding original continuous waveform data are input into the Transformer encoder. The authenticity of the event is determined by the self-attention mechanism to obtain the real microseismic event. The P-wave and S-wave first arrival times of the real microseismic event are picked up to obtain the P / S-wave first arrival times of each real microseismic event.

5. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 4, characterized in that, The specific process of the three-dimensional joint positioning includes: Based on the first arrival times of the P-wave and S-wave of each real microseismic event, the simplex method is used to locate each real microseismic event and obtain preliminary location results. Based on the waveform cross-correlation coefficients between the real microseismic events, the real microseismic events are associated and paired to obtain multiple event pairs; Construct double-difference localization equations for all event pairs and solve them to obtain the location correction parameters for each of the real microseismic events. The preliminary location results of the actual microseismic event are corrected by the location correction parameters to obtain the microseismic event location results.

6. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 5, characterized in that, The double-difference localization equation for each pair of events is: ； in, For events in an event pair and events Reach any observation point The travel time difference of P / S waves, where the observation point is any one of the ground station and the monitoring well; For the event Arrive at the observation point The travel time of the P / S wave; For the event Arrive at the observation point P / S wave travel time; It is an event and events Spatial location difference, It is an event and events The time difference of the earthquake's occurrence; For the event The travel-time partial derivative; For the event The traveling partial derivatives.

7. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 1 or 6, characterized in that, The specific process of the focal mechanism inversion includes: Based on the microseismic event location results, valid microseismic events with a sufficient number of effective records at observation points and complete waveform data were selected. Based on the location results of the effective microseismic events, six sets of basic Green's functions corresponding to each observation point of the effective microseismic event are matched from a pre-constructed three-dimensional Green's function database. The six sets of basic Green's functions correspond one-to-one with the six independent components of the moment tensor. For each effective microseismic event, the theoretical simulated waveform is obtained by convolving each set of basic Green's functions with the corresponding moment tensor components. With the goal of minimizing the L2 norm residual between the measured waveform and the theoretical simulated waveform, the optimal moment tensor for each effective microseismic event is obtained by iterative optimization using the adjoint state method combined with the L-BFGS algorithm. The optimal moment tensor is orthogonally decomposed to obtain the source rupture characteristic parameters of each effective microseismic event; The three-dimensional Green's function database is constructed according to the following steps: the study area of ​​the three-dimensional geomechanical model is divided into grids according to space and depth to obtain multiple grid nodes; based on the three-dimensional velocity model, the spectral element method is used to calculate the six sets of basic Green's functions corresponding to each observation point of each grid node; and the Green's functions of all grid nodes are collected to form a three-dimensional Green's function database covering the study area.

8. The gas storage safety monitoring method based on well-ground joint observation optimization according to claim 7, characterized in that, The specific process of multi-scale inversion of the regional stress field includes: Based on the microseismic event location results, the effective microseismic events are divided into three time series subsets: the pre-pressurization stage, the rapid pressurization stage, and the pressure stabilization stage. Based on the Wallace-Bott mechanical assumptions, and constrained by the source rupture characteristic parameters of microseismic events within each time series subset, the stress tensor of the corresponding region in the corresponding time period is inverted using the Bayesian MCMC method to obtain the stress field inversion results for each time series stage. The mechanical parameter inversion results include the stress field inversion results and the source rupture characteristic parameters.

9. A gas storage safety monitoring device based on well-ground joint observation optimization, characterized in that, The device includes: The signal acquisition unit is used to acquire physical field monitoring data of the target gas storage facility based on the three-dimensional observation network constructed at the target gas storage facility. The detection and positioning unit is used to perform microseismic event detection and three-dimensional joint positioning on the physical field monitoring data to obtain the microseismic event positioning results. The data inversion unit is used to perform source mechanism inversion and regional stress field multi-scale inversion based on the microseismic event location results and the physical field monitoring data, and to obtain mechanical parameter inversion results. The assessment and early warning unit is used to perform dynamic risk assessment and graded early warning based on the microseismic event location results, the mechanical parameter inversion results, and the physical field monitoring data.

10. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program stored in the memory, implements the gas storage safety monitoring method based on well-ground joint observation optimization as described in any one of claims 1 to 8.

Images