Safety blowout control method for deepwater high-yield gas well ground flow

Abstract

The invention discloses a deepwater high-yield gas well ground flow safe blowout control method, which belongs to the technical field of petroleum and natural gas exploration and development, and comprises the following steps: determining a weight coefficient set comprising noisy vibration signals, frequency domain characteristics and weight coefficients of fluid flow velocity data relative to pipeline pressure data; calculating and predicting a pressure change sequence according to the fluid flow rate data, the de-noised vibration signal, the frequency domain characteristic data, the pipeline pressure data and the weight coefficient set; according to the pressure change sequence and the target pressure range, a pressure deviation value and a deviation change rate are obtained, and a valve opening degree adjusting instruction grade is obtained according to the pressure deviation value and the deviation change rate; and an updated valve opening value is obtained according to the valve opening adjusting instruction grade, the current pipeline pressure data and the current valve opening value. According to the safety blowout control method for the ground flow of the deepwater high-yield gas well, the problem that an existing valve control mode is not beneficial to oil gas testing of the ultra-deepwater high-yield gas well is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and natural gas exploration and development, and in particular to a surface process safety blowout control method for a deepwater high-yield gas well. Background Art

[0002] As a key pillar of the national economy, oil and gas extraction efficiency and safety are directly linked to energy supply security and economic stability. During oil and gas testing in ultra-deepwater, high-yield gas wells, oil and gas flowing from the bottom of the well to the surface (offshore platform) often experience high pressure and velocity, posing significant safety threats to the surface testing process, such as vibration, excessive flow rates, and pressure. Current surface testing risk control methods primarily rely on emergency shutdown systems (ESDs) to control surface safety valves. When a hazard is detected, emergency shutdown buttons distributed throughout the testing process are manually triggered to close the safety valves, shutting down the well and controlling the hazard. However, ESDs are primarily used to control critical hazard conditions. Once triggered, the well must be shut down, interrupting testing, impacting the recording of test data, and negatively affecting subsequent reservoir evaluation. Therefore, current control methods struggle to develop targeted valve control mechanisms for vibration, excessive flow rates, and pressure, hindering testing. Summary of the Invention

[0003] In order to overcome the defects of the prior art, the present invention provides a method for controlling the surface process safety blowout of a deepwater high-yield gas well to solve the above-mentioned problems.

[0004] The technical solution adopted by the present invention to solve the technical problem is: a method for controlling the surface process safety blowout of a deepwater high-yield gas well, comprising the following steps:

[0005] S1: Obtain the original vibration value data of the pipeline; filter the noise of the original vibration value data to obtain a denoised vibration signal, and perform feature extraction processing on the denoised vibration signal to obtain frequency domain features;

[0006] S2: Obtain fluid flow rate data and pipeline pressure data; determine a set of weight coefficients including weight coefficients of the noisy vibration signal, frequency domain characteristics, and fluid flow rate data relative to the pipeline pressure data;

[0007] S3: The pressure prediction module calculates the fluid velocity data, denoised vibration signal, frequency domain feature data, pipeline pressure data and weight coefficient set at the current moment and historical moments to obtain the pressure change sequence within the future preset time window;

[0008] S4: Compare and calculate the pressure variation sequence within the preset time window with the target pressure range preset by the system to obtain the pressure deviation and the deviation change rate. The pressure deviation and the deviation change rate are used as inputs to the pressure regulation control rule library to output the valve opening adjustment instruction level.

[0009] S5: Obtaining a valve opening adjustment amount according to the valve opening adjustment instruction level and the current pipeline pressure data; and obtaining an updated valve opening value by combining the current valve opening value and the valve opening adjustment amount.

[0010] Preferably, in step S1, original vibration value data collected by the vibration sensor is obtained; the original vibration value data is subjected to noise filtering using a least mean square algorithm to generate a denoised vibration signal;

[0011] The denoised vibration signal is analyzed in the frequency domain by fast Fourier transform to obtain the frequency domain feature data.

[0012] Optionally, in step S2, the fluid flow rate data, pipeline pressure data, denoised vibration signal and frequency domain feature data are aligned in time series to construct a four-dimensional input data matrix; and a set of weight coefficients including the weight coefficients of the noisy vibration signal, frequency domain features and fluid flow rate data relative to the pipeline pressure data are determined through correlation analysis.

[0013] It is worth noting that in step S2, based on the four-dimensional input data matrix, the Pearson correlation coefficient formula is used. Calculate the correlation value r of each variable relative to the pipeline pressure data jp , where x ij is the pressure data of the i-th pipeline, represents the mean value of pipeline pressure data, p i is the i-th variable, which is one of the fluid velocity data, denoised vibration signal and frequency domain feature data. Represents the mean of the corresponding variable; the correlation value r jp After normalization, it is used as the weight coefficient.

[0014] Specifically, in step S3, the vibration amplitude is obtained by denoising the vibration signal, the amplitude change rate is calculated based on the vibration amplitude, and the flow rate fluctuation coefficient is calculated based on the fluid flow rate data; if the amplitude change rate exceeds a preset threshold and the flow rate fluctuation coefficient is greater than a critical value, an unstable system state identification signal is output; and the pressure prediction module is started according to the unstable system state identification signal.

[0015] Preferably, in step S3, the weight coefficient set includes a weight coefficient of the fluid velocity data relative to the pipeline pressure data, a weight coefficient of the denoised vibration signal relative to the pipeline pressure data, and a weight coefficient of the frequency domain feature data relative to the pipeline pressure data;

[0016] According to the fluid flow rate data, denoised vibration signal, frequency domain feature data and corresponding weight coefficients in the weight coefficient set at the current moment and the historical moment, the weighted fluid flow rate data, the weighted denoised vibration signal and the weighted frequency domain feature data at the corresponding moment are obtained, and a time series is constructed according to the weighted fluid flow rate data, the weighted denoised vibration signal, the weighted frequency domain feature data and the pipeline pressure data;

[0017] The autoregressive integrated moving average tool is used to fit the time series and calculate the pressure change series including the predicted values ​​of pressure changes within a preset time window in the future.

[0018] Optionally, in step S4, the pressure change prediction value P at time t is obtained from the pressure change sequence. t and the preset target pressure range, select the median of the preset target pressure range as the pressure change target value P0; calculate the pressure deviation D corresponding to time t t , where D t =P t -P0;

[0019] Calculate the deviation change rate R between time t and the adjacent time t-1 t ,in Δt is the time interval between two adjacent predicted pressure values ​​in the pressure change sequence;

[0020] According to the pressure deviation D t The interval and the rate of change of deviation R t The interval is mapped to a preset rule input variable set to obtain corresponding input variables, and based on the input variables, the valve opening adjustment instruction level L corresponding to the time t is obtained in the pre-established pressure regulation control rule library.

[0021] Specifically, in step S5, the current pipeline pressure data P is obtained, and the valve opening adjustment amount ΔV=k*L(P-P0) is calculated according to the valve opening adjustment instruction level L and the current pipeline pressure data P, where k is a preset proportional coefficient;

[0022] The valve opening adjustment amount ΔV is added to the current valve opening value V0 to generate an updated valve opening value V=V0+ΔV.

[0023] It is worth noting that in step S5, if the updated valve opening value V exceeds the preset valve opening range, the formula V1 = maximum value (Vmin, minimum value (Vmax, V)) is used to obtain the final updated valve opening value V1; where Vmin and Vmax are the lower and upper boundaries of the preset range respectively.

[0024] The beneficial effects of the present invention are as follows: in the surface process safety blowout control method of deep-water high-yield gas wells, in order to solve the problem of system instability caused by vibration, flow rate and pressure exceeding the standard during the operation of gas well pipelines, pipeline vibration data is collected, and noise-removed and frequency domain features are extracted to obtain denoised vibration signals and frequency domain features. The weight coefficients of various parameters relative to pipeline pressure data are determined by combining the fluid flow rate and pressure data collected by the flow meter and pressure sensor. The present invention generates a pressure change sequence within a preset time window in the future through a pressure prediction module. Based on the pressure deviation and the deviation change rate, combined with the valve opening adjustment rule library, the valve opening adjustment instruction level is output, and finally the valve opening value is obtained to control the opening of the surface safety valve, thereby avoiding the situation in the prior art where the valve is directly shut off in the middle of the test, thereby facilitating the test and improving the efficiency of the test. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a flow chart of a method for controlling surface flow safety in a deepwater high-yield gas well according to one embodiment of the present invention;

[0026] Figure 2 1 is a step-by-step flow chart of step S3 in one embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following is a further description of specific embodiments of the present invention in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is intended to facilitate understanding of the present invention and does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0028] like Figure 1-2 As shown, a method for controlling surface process safety blowout of a deepwater high-yield gas well comprises the following steps:

[0029] S1: Obtain the original vibration value data of the pipeline; filter the noise of the original vibration value data to obtain a denoised vibration signal, and perform feature extraction processing on the denoised vibration signal to obtain frequency domain features;

[0030] S2: Obtain fluid flow rate data measured and collected by the flow meter and pipeline pressure data collected by the pressure sensor; determine a set of weight coefficients including weight coefficients of the noisy vibration signal, the frequency domain characteristics, and the fluid flow rate data relative to the pipeline pressure data;

[0031] S3: The pressure prediction module calculates the fluid velocity data, denoised vibration signal, frequency domain feature data, pipeline pressure data and weight coefficient set at the current moment and historical moments to obtain the pressure change sequence within the future preset time window;

[0032] S4: Compare and calculate the pressure variation sequence within the preset time window with the target pressure range preset by the system to obtain the pressure deviation and the deviation change rate. The pressure deviation and the deviation change rate are used as inputs to the pressure regulation control rule library to output the valve opening adjustment instruction level.

[0033] S5: Obtaining a valve opening adjustment amount according to the valve opening adjustment instruction level and the current pipeline pressure data; and obtaining an updated valve opening value by combining the current valve opening value and the valve opening adjustment amount.

[0034] In the method for controlling the safe blowout of the surface process of deep-water high-yield gas wells, the problem of system instability caused by vibration, flow rate and pressure exceeding the standard during the operation of the gas well pipeline is addressed. By collecting pipeline vibration data, noise filtering and frequency domain feature extraction, a denoised vibration signal and frequency domain features are obtained. The weight coefficient of each parameter relative to the pipeline pressure data is determined by combining the fluid flow rate and pressure data collected by the flow meter and pressure sensor. The present invention generates a pressure change sequence within a preset time window in the future through a pressure prediction module. Based on the pressure deviation and the deviation change rate, combined with the valve opening adjustment rule library, the valve opening adjustment instruction level is output, and finally the valve opening value is obtained to control the opening of the ground safety valve, thereby avoiding the situation in the existing technology where the valve is directly shut off in the middle of the test, which is beneficial to the progress of the test and improves the efficiency of the test.

[0035] It is worth noting that in step S1, the original vibration value data collected by the vibration sensor is obtained; the original vibration value data is subjected to noise filtering using the least mean square algorithm to generate a denoised vibration signal;

[0036] The denoised vibration signal is analyzed in the frequency domain by fast Fourier transform to obtain the frequency domain feature data.

[0037] The vibration sensor collects vibration data at a high frequency of 1000 times per second, generating raw vibration data that includes pipeline vibration and ambient noise. Ambient noise can include nearby mechanical vibration or electromagnetic interference, resulting in a significant amount of irrelevant components in the signal, affecting the accuracy of subsequent analysis. To filter out noise, a least mean square (LMS) algorithm is used to process the raw vibration data. Initializing the filter weights to zero vectors means the filter begins learning signal characteristics from an unbiased state. A preset step size of 0.01 ensures a smooth weight update process and avoids convergence instability caused by excessively large step sizes. For example, when processing railway vibration signals, the desired signal can be set to an idealized track vibration waveform collected in a laboratory, and the output error is the difference between the actual signal and the desired signal. With each iteration, the filter adjusts the weights based on the error, gradually approximating the desired signal until the error falls below a preset threshold, such as 0.001. If the error is too high, significant noise will remain. Through iteration, the resulting denoised vibration signal effectively filters out nearby mechanical vibration or electromagnetic interference, preserving the key characteristics of pipeline vibration. This denoising method significantly improves the signal-to-noise ratio, providing a clearer data foundation for subsequent analysis. Specifically, a Fast Fourier Transform (FFT) is performed on the denoised vibration signal to extract frequency domain feature data for analysis of vibration characteristics. It should be noted that the frequency resolution of the FFT is related to the sampling time length: longer sampling times yield higher resolution, but also increase computational complexity. In practical applications, one-second signal segments are typically selected for the transformation to balance accuracy and efficiency. The high signal-to-noise ratio of the denoised vibration signal ensures the accuracy of the frequency domain analysis, and the frequency domain feature data provides a quantitative basis for fault diagnosis.

[0038] Preferably, in step S2, the fluid flow rate data, pipeline pressure data, denoised vibration signal and frequency domain feature data are aligned in time series to construct a four-dimensional input data matrix; and a set of weight coefficients including the weight coefficients of the noisy vibration signal, frequency domain features and fluid flow rate data relative to the pipeline pressure data are determined by correlation analysis.

[0039] Optionally, in step S2, based on the four-dimensional input data matrix, the Pearson correlation coefficient formula is used Calculate the correlation value r of each variable relative to the pipeline pressure data jp , where x ij is the pressure data of the i-th pipeline, represents the mean value of pipeline pressure data, p i is the i-th variable, which is one of the fluid velocity data, denoised vibration signal and frequency domain feature data. Represents the mean of the corresponding variable; the correlation value r jp After normalization, it is used as the weight coefficient.

[0040] In one possible implementation, the acquisition of fluid velocity data measured by a flow meter and pipeline pressure data collected by a pressure sensor must be synchronized with the vibration data collection to achieve a unified timestamp and sampling frequency. Preferably, the flow meter uses an ultrasonic flow meter, which calculates flow velocity using the time difference of sound wave propagation, with an accuracy of up to 0.01 cubic meters per second. The pressure sensor uses a high-precision capacitive sensor with a measurement range of 0-10 MPa and an accuracy of 0.1%. This high-precision equipment ensures the reliability of the raw data, laying the foundation for subsequent data alignment and analysis.

[0041] In this embodiment, time alignment uses linear interpolation to fill in small deviations in timestamps, ensuring that each data point in the matrix accurately corresponds to the time axis. For example, at time t = 1.000 seconds, the flow rate is 2.5 cubic meters per second, the pressure is 5.2 MPa, the amplitude of the denoised vibration signal is 0.03 meters per second squared, and the frequency domain characteristics show a dominant frequency of 20 Hz. These data constitute a row of the matrix.

[0042] In one possible implementation, the correlation between each parameter and the pipeline pressure data is calculated based on a four-dimensional input data matrix, using the Pearson correlation coefficient formula. For example, a one-minute data segment containing 6,000 sampling points is selected to calculate the correlation between flow rate and pressure. The flow rate data fluctuates between 2.0 and 3.0 cubic meters per second, and the pressure data range is 5.0 to 5.5 MPa. Through mean calculation and normalization processing, the correlation coefficient between flow rate and pressure is 0.85. Similarly, the correlation coefficient between the denoised vibration signal and pressure is 0.65, and the correlation between the main frequency of the frequency domain feature and pressure is 0.45. After normalization, these correlation coefficients can be used as weight coefficients and thus combined into a set of weight coefficients.

[0043] Specifically, in step S3, the vibration amplitude is obtained by denoising the vibration signal, the amplitude change rate is calculated based on the vibration amplitude, and the flow rate fluctuation coefficient is calculated based on the fluid flow rate data; if the amplitude change rate exceeds a preset threshold and the flow rate fluctuation coefficient is greater than a critical value, an unstable system state identification signal is output; and the pressure prediction module is started according to the unstable system state identification signal.

[0044] Extract the vibration amplitude of each frequency component from the frequency domain feature data and form a vibration amplitude sequence based on the time series. According to the vibration amplitude sequence, the adjacent data point difference calculation formula is used Among them A i For the i-th amplitude data point, the i-th amplitude change rate Δ is calculated i , all amplitude change rates are combined into an amplitude change rate sequence. The standard deviation calculation formula is used for the velocity sequence formed by the time series of fluid velocity data where v tis the tth flow velocity value, is the average flow rate, T is the total number of data points, and the calculated standard deviation σ v As the flow rate fluctuation coefficient. The preset threshold corresponding to the amplitude change rate is set to 15%. If the amplitude change rate exceeds the preset threshold, it is marked as an amplitude exceeding limit state. The critical value of the flow rate fluctuation coefficient is set to 0.2. If the flow rate fluctuation coefficient exceeds the critical value, it is marked as a fluctuation exceeding limit state. A logical AND operation is used to determine the amplitude exceeding limit state and the fluctuation exceeding limit state. When both states are met simultaneously within the set time sliding window, an unstable system state identification signal is generated.

[0045] It is worth noting that in step S3, the weight coefficient set includes the weight coefficient of the fluid velocity data relative to the pipeline pressure data, the weight coefficient of the denoised vibration signal relative to the pipeline pressure data, and the weight coefficient of the frequency domain feature data relative to the pipeline pressure data;

[0046] According to the fluid flow rate data, denoised vibration signal, frequency domain feature data and corresponding weight coefficients in the weight coefficient set at the current moment and the historical moment, the weighted fluid flow rate data, the weighted denoised vibration signal and the weighted frequency domain feature data at the corresponding moment are obtained, and a time series is constructed according to the weighted fluid flow rate data, the weighted denoised vibration signal, the weighted frequency domain feature data and the pipeline pressure data;

[0047] The autoregressive integrated moving average tool is used to fit the time series and calculate the pressure change series including the predicted values ​​of pressure changes within a preset time window in the future.

[0048] Specifically, the Autoregressive Integrated Moving Average (ARIMA) tool is used to fit time series and predict future pressure changes. The ARIMA model determines model parameters by analyzing the stationarity and autocorrelation of historical pressure data. For example, in a pipeline scenario, a preset time window is set to 10 minutes, and pressure data is collected every minute. Historical data shows that the pressure fluctuation period is 5 minutes. The ARIMA model can fit this trend and predict the pressure value in the next 10 minutes, such as a change from 5.0 MPa to 5.2 MPa. In one embodiment, the predicted pressure change value for the preset time window in the future is calculated to generate a pressure change sequence. For example, within a 10-minute window, the predicted value at the first moment is 5.0 MPa, the predicted value at the second moment is 5.2 MPa, and the predicted value at the third moment is 4.9 MPa, forming a sequence. This sequence can be used to identify potential pressure anomalies. It is understood that the appropriate setting of weight coefficients and the application of the ARIMA model can improve the accuracy of the predicted sequence. Preferably, the generation of the pressure change sequence incorporates the field background of vibration amplitude and flow velocity fluctuation analysis mentioned in the historical conversation.

[0049] Preferably, in step S4, the pressure change prediction value P at time t is obtained from the pressure change sequence. t and the preset target pressure range, select the median of the preset target pressure range as the pressure change target value P0; calculate the pressure deviation D corresponding to time t t , where D t =P t -P0, get the pressure deviation sequence;

[0050] Calculate the deviation change rate R between time t and the adjacent time t-1 t ,in Δt is the time interval between two adjacent predicted pressure values ​​in the pressure change sequence, and the deviation change rate sequence is obtained;

[0051] According to the pressure deviation D t The interval and the rate of change of deviation R t The interval is mapped to a preset rule input variable set, such as the input variable set {negative large, negative small, zero, positive small, positive large}, to obtain the corresponding input variable, and based on the input variable, the valve opening adjustment instruction level L corresponding to the time t is obtained in the pre-established pressure regulation control rule library.

[0052] Pressure deviation D t It reflects the degree of deviation between the predicted value and the target value, and the deviation change rate R t The trend of deviation change over time is captured. Assume that the preset target pressure range of a natural gas pipeline is 4.8 to 5.2 MPa, and the median value 5.0 MPa is selected as the pressure change target value P0. If the pressure change prediction value P at a certain time t is t is 5.3 MPa, then D t is 0.3MPa. If the time interval Δt is 1 minute, the previous moment D t-1 is 0.2MPa, then R t This calculation method can quantify the dynamic characteristics of pressure fluctuations and provide a data basis for subsequent calculations.

[0053] In a possible implementation, the pressure deviation D t and the rate of change of deviation R t Is mapped to the rule input variable set {negative large, negative small, zero, positive small, positive large}. For example, the pressure deviation D t The interval can be set as: D t <-0.2Mpa is negative, -0.2≤D t <-0.05Mpa is negative, -0.05≤D t ≤-0.05Mpa is zero, 0.05<D t ≤0.2Mpa is positive small, Dt >0.2Mpa is positive. Deviation change rate R t The interval is similar, such as R t <-0.1Mpa / min is negative, R t >0.1Mpa / min is positive. Assume that the pressure deviation D at a certain moment t is 0.3MPa, the deviation change rate R t The pressure deviation after mapping is 0.15MPa / min. t is positive, the deviation change rate R t This mapping method converts continuous values ​​into discrete rule inputs, which is convenient for matching with the pressure regulation control rule base. Specifically, the pressure regulation control rule base is built based on historical data and expert experience, and contains the corresponding relationship between input variables and valve opening adjustment instruction level L. For example, the rule base may stipulate: When the pressure deviation D t is positive and the deviation change rate R t When the pressure deviation D is positive, L=5; t is zero and the deviation change rate R t When it is a negative hour, L=2.

[0054] Optionally, in step S5, current pipeline pressure data P is obtained, and the valve opening adjustment amount ΔV=k*L(P-P0) is calculated according to the valve opening adjustment instruction level L and the current pipeline pressure data P, where k is a preset proportional coefficient;

[0055] The valve opening adjustment amount ΔV is added to the current valve opening value V0 to generate an updated valve opening value V=V0+ΔV.

[0056] For example, if the valve opening adjustment command level L = 2, the proportional coefficient k is preset to 0.05, the current pipeline pressure data P is 5.2 MPa, the target pressure change value P0 is 5 MPa, and the valve opening adjustment value ΔV is calculated, ΔV is 0.02. This method converts the pressure deviation into a specific opening adjustment value by introducing a proportional coefficient, ensuring that the adjustment action is proportional to the degree of deviation, which is suitable for dynamic operating conditions. Specifically, when calculating the updated valve opening value V, it is based on the current valve opening value V0. For example, if the current valve opening value V0 is 0.6, then V = V0 + ΔV = 0.6 + 0.02 = 0.62. This step generates a new control target value by accumulating the adjustment value, maintaining the continuity of the valve action.

[0057] Specifically, in step S5, if the updated valve opening value V exceeds the preset valve opening range, it is limited to the range boundary, and the formula V1 = maximum value (Vmin, minimum value (Vmax, V)) is used to obtain the final updated valve opening value V1; wherein Vmin and Vmax are the lower boundary and upper boundary of the preset range, respectively.

[0058] In one embodiment, if the updated valve opening value V exceeds the preset valve opening range, for example, Vmin is 0 and Vmax is 1, its value is limited. Assuming the updated valve opening value V is calculated to be 1.05, then V1 = maximum value (Vmin, minimum value (Vmax, V)) = 1. If the updated valve opening value V is -1%, then V1 = 0. This limiting mechanism ensures that the valve opening remains within the physically feasible range, preventing overshoot or ineffective regulation.

[0059] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. It is apparent to those skilled in the art that various changes, modifications, substitutions, and variations to these embodiments may be made without departing from the principles and spirit of the present invention, and these changes and modifications still fall within the scope of protection of the present invention.

Claims

1. A method for controlling surface flow safety of deepwater high-yield gas wells, characterized in that: The following steps are involved: S1: Obtain the original vibration value data of the pipeline; filter the noise of the original vibration value data to obtain a denoised vibration signal, and perform feature extraction processing on the denoised vibration signal to obtain frequency domain features; S2: Obtain fluid flow rate data and pipeline pressure data; determine a set of weight coefficients including weight coefficients of the noisy vibration signal, frequency domain characteristics, and fluid flow rate data relative to the pipeline pressure data; S3: The pressure prediction module calculates the fluid velocity data, denoised vibration signal, frequency domain feature data, pipeline pressure data and weight coefficient set at the current moment and historical moments to obtain the pressure change sequence within the future preset time window; S4: Compare and calculate the pressure variation sequence within the preset time window with the target pressure range preset by the system to obtain the pressure deviation and the deviation change rate. The pressure deviation and the deviation change rate are used as inputs to the pressure regulation control rule library to output the valve opening adjustment instruction level. S5: Obtaining a valve opening adjustment amount according to the valve opening adjustment instruction level and the current pipeline pressure data; and obtaining an updated valve opening value by combining the current valve opening value and the valve opening adjustment amount.

2. A method for controlling surface flow safety of a deepwater high-yield gas well according to claim 1, characterized in that: In step S1, the original vibration value data collected by the vibration sensor is obtained; the original vibration value data is subjected to noise filtering using a least mean square algorithm to generate a denoised vibration signal; The denoised vibration signal is analyzed in the frequency domain by fast Fourier transform to obtain the frequency domain feature data.

3. The method for controlling surface flow safety of a deepwater high-yield gas well according to claim 1, characterized in that: In step S2, the fluid flow rate data, pipeline pressure data, denoised vibration signal and frequency domain feature data are aligned in time series to construct a four-dimensional input data matrix; and a set of weight coefficients including the weight coefficients of the noisy vibration signal, frequency domain feature and fluid flow rate data relative to the pipeline pressure data is determined through correlation analysis.

4. A method for controlling surface flow safety of a deepwater high-yield gas well according to claim 3, characterized in that: In step S2, based on the four-dimensional input data matrix, the Pearson correlation coefficient formula is used Calculate the correlation value r of each variable relative to the pipeline pressure data jp , where x ij is the pressure data of the i-th pipeline, represents the mean value of pipeline pressure data, p i is the i-th variable, which is one of the fluid velocity data, denoised vibration signal and frequency domain feature data. Represents the mean of the corresponding variable; the correlation value r jp After normalization, it is used as the weight coefficient.

5. The method for controlling surface process safety blowout of a deepwater high-yield gas well according to claim 1, characterized in that: In step S3, the vibration amplitude is obtained by denoising the vibration signal, the amplitude change rate is calculated based on the vibration amplitude, and the flow rate fluctuation coefficient is calculated based on the fluid flow rate data; if the amplitude change rate exceeds a preset threshold and the flow rate fluctuation coefficient is greater than a critical value, an unstable system state identification signal is output; and the pressure prediction module is started according to the unstable system state identification signal.

6. The method for controlling surface flow safety of a deepwater high-yield gas well according to claim 1, characterized in that: In step S3, the weight coefficient set includes the weight coefficient of the fluid velocity data relative to the pipeline pressure data, the weight coefficient of the denoised vibration signal relative to the pipeline pressure data, and the weight coefficient of the frequency domain feature data relative to the pipeline pressure data; According to the fluid flow rate data, denoised vibration signal, frequency domain feature data and corresponding weight coefficients in the weight coefficient set at the current moment and the historical moment, the weighted fluid flow rate data, the weighted denoised vibration signal and the weighted frequency domain feature data at the corresponding moment are obtained, and a time series is constructed according to the weighted fluid flow rate data, the weighted denoised vibration signal, the weighted frequency domain feature data and the pipeline pressure data; The autoregressive integrated moving average tool is used to fit the time series and calculate the pressure change series including the predicted values ​​of pressure changes within a preset time window in the future.

7. The method for controlling surface flow safety of a deepwater high-yield gas well according to claim 1, characterized in that: In step S4, the pressure change prediction value P at time t is obtained from the pressure change sequence. t and a preset target pressure range, selecting the median of the preset target pressure range as the pressure change target value P0; Calculate the pressure deviation D corresponding to time t t , where D t =P t -P0; Calculate the deviation change rate R between time t and the adjacent time t-1 t ,in Δt is the time interval between two adjacent predicted pressure values ​​in the pressure change sequence; According to the pressure deviation D t The interval and the rate of change of deviation R t The interval is mapped to a preset rule input variable set to obtain corresponding input variables, and based on the input variables, the valve opening adjustment instruction level L corresponding to the time t is obtained in the pre-established pressure regulation control rule library.

8. The method for controlling surface flow safety of a deepwater high-yield gas well according to claim 1, characterized in that: In step S5, the current pipeline pressure data P is obtained, and the valve opening adjustment amount ΔV=k*L(P-P0) is calculated according to the valve opening adjustment instruction level L and the current pipeline pressure data P, where k is a preset proportional coefficient; The valve opening adjustment amount ΔV is added to the current valve opening value V0 to generate an updated valve opening value V=V0+ΔV.

9. A method for controlling surface flow safety of a deepwater high-yield gas well according to claim 8, characterized in that: In step S5, if the updated valve opening value V exceeds the preset valve opening range, the formula V1 = maximum value (Vmin, minimum value (Vmax, V)) is used to obtain the final updated valve opening value V1; where Vmin and Vmax are the lower and upper boundaries of the preset range respectively.

Images