Injection-production induced earthquake risk management and control method and system based on multi-source data fusion

Abstract

The invention relates to the field of seismic data processing, and particularly discloses an injection-production induced seismic risk management and control method and system based on multi-source data fusion, and the method comprises the steps: collecting multi-source data, processing the multi-source data, extracting a feature vector subset of a key variable of an induced seismic event, and carrying out the multi-source data fusion. Extracting prediction data vectors corresponding to key variable characteristics of an induced earthquake event, constructing and training an override probability model of dynamic risk assessment, establishing a mapping relation between multiple indexes and real-time risks, establishing a real-time grading early warning and feedback management and control mechanism, and judging whether the real-time grading early warning and feedback management and control mechanism is triggered or not according to the earthquake magnitude of field operation; if yes, engineering parameters are dynamically adjusted according to the real-time risk, and closed-loop management and control are formed. According to the method, real-time assessment of the earthquake risk induced by injection-production operation can be realized, and the method is effectively combined with field construction process management and control, so that the effect of resisting the earthquake risk induced by industrial production activities is effectively improved.

Description

Technical Field

[0001] The present application relates to the field of seismic data processing, and in particular to a method and system for controlling injection-production induced earthquake risks based on multi-source data fusion. Background Art

[0002] In recent years, with the increase of related industrial activities (such as hydraulic fracturing, wastewater reinjection, geothermal development, etc.), induced earthquake events have occurred frequently, posing a threat to engineering safety, ecological environment and public property. Therefore, the prediction of induced earthquakes and the reasonable control of earthquake magnitude are particularly critical for the smooth progress of on-site industrial activities.

[0003] The "traffic light" system is a powerful decision-making tool for managing industrial activities. It has been widely used to control induced seismic events related to gas and liquid injection and production operations, and to guide government departments and construction companies to take appropriate response measures.

[0004] In this risk management system, earthquake monitoring methods are particularly important. However, traditional earthquake monitoring methods rely on data from a single seismic network, which often suffers from isolation and lag. Specifically, data on earthquakes, geology, and engineering activities are dispersed and lack multi-dimensional fusion analysis. Furthermore, static models based on historical statistics cannot adapt to the real-time risk changes in dynamic construction scenarios. Furthermore, traditional prediction methods are mostly driven by physical models. However, a single physical model or empirical formula cannot quantify the critical threshold for induced earthquakes under complex geological and engineering conditions, making it difficult to achieve early warning and precise risk management. Summary of the Invention

[0005] In order to improve the problem that existing technologies are difficult to achieve early warning and precise control of induced earthquake risks, the present application provides a method and system for injection-production induced earthquake risk control based on multi-source data fusion.

[0006] This application provides a method and system for managing and controlling injection-production induced earthquake risks based on multi-source data fusion, which adopts the following technical solutions: A method and system for controlling injection-production induced earthquake risk based on multi-source data fusion, comprising the following steps: Collect multi-source data, including seismic network monitoring data, geological survey data, construction engineering data, and environmental monitoring data; Processing the multi-source data to extract a subset of eigenvectors of key variables of induced earthquake events; Based on the feature vector subset of the key variables of the induced earthquake event, multi-source data fusion is performed to extract the prediction data vector corresponding to the key variable characteristics of the induced earthquake event; Build and train a exceedance probability model for dynamic risk assessment, input the predicted data vector corresponding to the key variable characteristics of induced earthquake events, output the daily induced earthquake occurrence rate, i.e., real-time risk, and establish a mapping relationship between multiple indicators and real-time risk; Quantify seismicity parameters, geological sensitivity factors, and engineering disturbance intensity as risk indicators; Establish a real-time hierarchical early warning and feedback control mechanism based on the real-time risks and risk indicators; The magnitude of the earthquake at the scene determines whether the real-time graded warning and feedback control mechanism is triggered; if triggered, the engineering parameters are adjusted dynamically according to the real-time risk to form a closed-loop control.

[0007] Furthermore, the seismic network monitoring data includes microseismic event waveforms, magnitudes, and focal mechanism solutions; geological survey data includes fault distribution, ground stress field, and lithologic parameters; construction engineering data includes injection and production fluid pressure / flow, fracturing fluid volume, and injection and production methods; and environmental monitoring data includes groundwater levels and surface deformation.

[0008] Furthermore, the multi-source data fusion is implemented using graph neural network fusion technology, and BP neural network is used for fault reasoning.

[0009] Furthermore, the exceedance probability model of the dynamic risk assessment is expressed as:

[0010] Where, is the cumulative volume of injected fluid, is the rate of injection of fluid, is a set of model parameters that describe the formation characteristics. It is activity feedback. is the earthquake magnitude ratio, is the average release time in days, It's time to shut down the well. M is the magnitude, M saf is the assumed safety magnitude, Y Represents the probability that the magnitude exceeds the threshold. When the magnitude of the induced earthquake exceeds the threshold, Y =1.

[0011] Furthermore, the daily induced earthquake occurrence rate is expressed as:

[0012] Where, t is the injection time in days, M 0 is the minimum cutoff magnitude.

[0013] Furthermore, the mapping relationship between the multiple indicators and real-time risks is established through a random forest regression model.

[0014] Furthermore, the seismic activity parameters include b-value and seismic energy release rate, the geological sensitivity factors include fault activation probability and ground stress accumulation, and the engineering disturbance intensity includes injection volume and pressure change gradient.

[0015] Furthermore, the real-time graded warning and feedback control mechanism includes: continuous monitoring under low risk, reducing the injection rate and optimizing the construction plan under medium risk, and immediately suspending operations and initiating emergency response under high risk; the degree of risk is determined by the set threshold magnitude and the actual magnitude on site.

[0016] Furthermore, the dynamically adjusted engineering parameters include an injection-production pressure threshold, an injection-production rate threshold, an injection-production pause time threshold, and a coordinated injection-production mode.

[0017] The present application also provides an injection-production induced earthquake risk management system based on multi-source data fusion, including a processor configured to execute a stored computer program to implement an injection-production induced earthquake risk management method based on multi-source data fusion.

[0018] In summary, this application has the following beneficial technical effects: This application uses deep fusion of multi-source heterogeneous data to break through the limitations of a single data dimension and improve the spatiotemporal resolution of risk identification; realize dynamic-static joint modeling, couple physical mechanism models with data-driven models, and enhance the generalization ability of complex scenarios; improve real-time feedback control, and establish an active prevention and control system of "construction, evaluation, and control" to reduce the probability of induced perceptible earthquakes in real time. In summary, this application realizes the real-time assessment of the risk of earthquakes induced by injection and production operations, and combines it with on-site construction process control to effectively improve the effectiveness of industrial production activities in resisting the risk of induced earthquakes. It is suitable for monitoring and controlling earthquake risks caused by industrial activities involving gas-liquid injection and production, such as wastewater reinjection, unconventional energy mining, geothermal development, and compressed air energy storage. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of a method for controlling injection-production induced earthquake risks based on multi-source data fusion according to an embodiment of the present application; Figure 2 This is a multi-source data fusion flow chart in an embodiment of the present application; Figure 3 This is a logic diagram of the hierarchical warning and feedback control mechanism in the embodiment of the present application; Figure 4 This is a diagram showing the effect of real-time control of earthquake magnitude in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The following is combined with Figure 1-4 This application is described in further detail.

[0021] The present application embodiment discloses a method for controlling the risk of injection-production induced earthquakes based on multi-source data fusion. Figure 1 and Figure 2 The risk management method for injection-production induced earthquakes based on multi-source data fusion includes the following steps: S1: Collect multi-source data, including seismic network monitoring data, geological survey data, construction engineering data, environmental monitoring data, and housing structure design data of residential areas around the site (if any).

[0022] Among them, the seismic network monitoring data includes microseismic event waveforms, magnitudes, and focal mechanism solutions; geological survey data includes fault distribution, ground stress field, and lithology parameters; construction engineering data includes injection and production fluid pressure / flow, fracturing fluid volume, and injection and production methods; and environmental monitoring data includes groundwater levels and surface deformation.

[0023] S2: Process multi-source data and extract the eigenvector subset of key variables that induce earthquake events.

[0024] Specifically, Kalman filtering is used in multi-source data processing to eliminate noise, which can be expressed as:

[0025] Where, is the prior state estimate, F is the state transition matrix, is the optimal estimate and posterior state estimate of the previous moment, B is the control input transfer matrix, is the control input, is the predicted covariance matrix, the prior covariance matrix, is the optimal estimate of the covariance matrix at the previous moment, the posterior covariance matrix, is the transpose of the state transfer matrix, Q is the process noise covariance matrix; is the Kalman gain coefficient, H To measure the transfer matrix, To measure the transpose of the transfer matrix, R is the measurement noise covariance matrix, is the optimal estimate, is the measured value, To predict the measured value, is the updated covariance matrix.

[0026] S3: Based on the feature vector subset of the key variables of the induced earthquake event, multi-source data fusion is performed to extract the prediction data vector corresponding to the key variable characteristics of the induced earthquake event.

[0027] Specifically, multi-source data fusion is implemented using graph neural network (GNN) fusion technology, and BP neural network is used for fault reasoning.

[0028] S4: Construct and train a exceedance probability model for dynamic risk assessment, input the predicted data vector corresponding to the key variable characteristics of induced earthquake events, output the daily induced earthquake occurrence rate, that is, real-time risk, and establish a mapping relationship between multiple indicators and real-time risk.

[0029] The specific method includes: converting the predicted data vector corresponding to the key variable characteristics of the induced earthquake event into the exceedance probability in combination with the safety standard. The exceedance probability model of dynamic risk assessment is expressed as:

[0030] Where, is the cumulative volume of injected fluid, is the rate of injection of fluid, is a set of model parameters that describe the formation characteristics. It is activity feedback. is the earthquake magnitude ratio, is the average release time in days, It's time to shut down the well. M is the magnitude, M saf is the assumed safety magnitude, Y Represents the probability that the magnitude exceeds the threshold. When the magnitude of the induced earthquake exceeds the threshold, Y =1.

[0031] In the hierarchical Bayesian framework, the probability of earthquake risk occurring due to the actual monitored magnitude exceeding the safe magnitude is obtained, and the daily induced earthquake occurrence rate is expressed as:

[0032] Where, t is the injection time in days, M 0 is the minimum cutoff magnitude.

[0033] Specifically, the mapping relationship between multiple indicators and real-time risks is established through the random forest regression model.

[0034] S5: Quantify seismicity parameters, geological sensitivity factors, and engineering disturbance intensity as risk indicators.

[0035] Among them, seismic activity parameters include b-value and seismic energy release rate, geological sensitivity factors include fault activation probability and ground stress accumulation, and engineering disturbance intensity includes injection volume and pressure change gradient.

[0036] S6: Establish a real-time hierarchical warning and feedback control mechanism based on real-time risks and risk indicators.

[0037] like Figure 3 As shown in the figure, the real-time graded warning and feedback control mechanism includes: continuous monitoring under low risk, reducing the injection rate and optimizing the construction plan under medium risk, and immediately suspending operations and initiating emergency response under high risk; the degree of risk is determined by the set threshold magnitude and the actual magnitude on site.

[0038] S7: Determine whether the real-time graded warning and feedback control mechanism is triggered based on the magnitude of the on-site operation; if triggered, adjust the engineering parameters dynamically according to the real-time risk to form a closed-loop control.

[0039] Among them, the dynamic adjustment of engineering parameters includes injection and production pressure threshold, injection and production rate threshold, injection and production suspension time threshold and coordinated injection and production method.

[0040] This example takes the earthquake induced by water fracturing in a double-well field as the research object and uses the above-mentioned injection-production-induced earthquake risk control method based on multi-source data fusion to control earthquake risk. Due to the need to exploit unconventional energy, two wells are used to inject water into the formation, with a total injection volume of 19440 m 3 The total mass is 23,328 tons, the construction period is initially set at 30 days, and the injection method is constant rate injection. Figure 3 As shown in the figure, when the magnitude of an earthquake detected by the local seismic network reaches level 3 during industrial production, magnitude control measures must be taken to trigger a real-time graded early warning and feedback control mechanism.

[0041] like Figure 4 As shown, after the sixth regulation, the magnitude of the injection-induced earthquake continues to decrease until it returns to the green light safety range. In contrast, the moment magnitude without safety regulation will continue to increase, proving that the regulation method proposed in this application can effectively reduce the risk of injection-induced earthquakes during on-site construction.

[0042] This application also provides a system for managing and controlling injection-production-induced seismic risk based on multi-source data fusion, including a processor configured to execute a stored computer program to implement the aforementioned method for managing and controlling injection-production-induced seismic risk based on multi-source data fusion. The system integrates three-dimensional geological modeling and risk heat maps, supports interactive analysis of multi-dimensional data, and provides an API interface for interaction with engineering control systems, building a visual operating platform for automated risk management.

[0043] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for managing and controlling injection-production induced seismic risk based on multi-source data fusion, characterized by: The following steps are involved: Collect multi-source data, including seismic network monitoring data, geological survey data, construction engineering data, and environmental monitoring data; Processing the multi-source data to extract a subset of eigenvectors of key variables of induced earthquake events; Based on the feature vector subset of the key variables of the induced earthquake event, multi-source data fusion is performed to extract the prediction data vector corresponding to the key variable characteristics of the induced earthquake event; Build and train a exceedance probability model for dynamic risk assessment, input the predicted data vector corresponding to the key variable characteristics of induced earthquake events, output the daily induced earthquake occurrence rate, i.e., real-time risk, and establish a mapping relationship between multiple indicators and real-time risk; Quantify seismicity parameters, geological sensitivity factors, and engineering disturbance intensity as risk indicators; Establish a real-time hierarchical early warning and feedback control mechanism based on the real-time risks and risk indicators; Determine whether to trigger the real-time graded warning and feedback control mechanism based on the magnitude of the on-site operation; If triggered, engineering parameters will be dynamically adjusted based on real-time risks to form a closed-loop control.

2. The method for managing and controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The seismic network monitoring data includes microseismic event waveforms, magnitudes, and focal mechanism solutions; geological survey data includes fault distribution, geostress field, and lithologic parameters; construction engineering data includes injection and production fluid pressure / flow, fracturing fluid volume, and injection and production methods; and environmental monitoring data includes groundwater levels and surface deformation.

3. The method for managing and controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The multi-source data fusion is achieved by adopting graph neural network fusion technology, and BP neural network is used for fault reasoning.

4. The method for controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The exceedance probability model of the dynamic risk assessment is expressed as: Where, is the cumulative volume of injected fluid, is the rate of injection of fluid, is a set of model parameters that describe the formation characteristics. It is activity feedback. is the earthquake magnitude ratio, is the average release time in days, It's time to shut down the well. M is the magnitude, M saf is the assumed safety magnitude, Y Represents the probability that the magnitude exceeds the threshold. When the magnitude of the induced earthquake exceeds the threshold, Y =1.

5. The method for managing and controlling injection-production induced seismic risk based on multi-source data fusion according to claim 4, characterized in that: The daily induced earthquake occurrence rate is expressed as: Where, t is the injection time in days, M 0 is the minimum cutoff magnitude.

6. The method for controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The mapping relationship between the multiple indicators and real-time risks is established through a random forest regression model.

7. The method for controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The seismic activity parameters include b value and seismic energy release rate, the geological sensitivity factors include fault activation probability and ground stress accumulation, and the engineering disturbance intensity includes injection volume and pressure change gradient.

8. The method for controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The real-time graded warning and feedback control mechanism includes: continuous monitoring under low risk, reducing the injection rate and optimizing the construction plan under medium risk, and immediately suspending operations and initiating emergency response under high risk; the degree of risk is determined by the set threshold magnitude and the actual magnitude on site.

9. The method for controlling injection-production induced seismic risk based on multi-source data fusion according to claim 1, characterized in that: The dynamically adjusted engineering parameters include an injection-production pressure threshold, an injection-production rate threshold, an injection-production pause time threshold, and a coordinated injection-production method.

10. A system for managing and controlling injection-production induced earthquake risk based on multi-source data fusion, comprising a processor, characterized in that: The processor is configured to execute a stored computer program to implement the method for managing injection-production induced seismic risks based on multi-source data fusion as described in any one of claims 1-9.

Images