A circulating gas lift pressurization parameter adaptive regulation method and system

Abstract

The application discloses a circulating gas lift pressurization parameter self-adaptive regulation and control method and system, and relates to the technical field of industrial control systems; its technical points are: obtaining circulating gas lift well operation data, establishing a candidate gas injection topology set and performing topology state constraints; in the reference recovery state, executing the positive sequence pressurization trajectory and the reverse sequence pressurization trajectory with the reverse pressurization order, compensating for the high-pressure gas path inventory mass, forming the reverse sequence differential response under the effective well cumulative mass coordinate, and performing continuous mass interval inspection with the reverse sequence differential reachable domain of each candidate gas injection topology; updating the topology conversion parameter set and the in-channel control parameter set according to conditional topology conversion evidence and fixed topology response, forming a topology control certificate after independent record verification; parallel propagation of the candidate annulus pressure reference sequence, screening of the common annulus pressure reference sequence and rolling execution, and application to the circulating gas lift control without continuous downhole valve position feedback.

Description

Technical Field

[0001] This invention relates to the field of industrial control system technology, specifically to an adaptive control method and system for circulating air lift boosting parameters. Background Technology

[0002] In circulating gas lift operations, the surface industrial control system controls the pressurization equipment and wellhead regulating valves to deliver high-pressure gas through the casing annulus to the downhole gas lift valve. Based on feedback from casing pressure, wellhead pressure, and injection gas flow rate, the system adjusts the target pressure, pressurization rate, and holding time. Downhole, unloading valves and working valves are typically installed at different depths. During the production unloading phase, gas is first injected through the shallower unloading valves. As annulus pressure builds up and the wellbore fluid level drops, the actual injection location gradually shifts to the deeper working valves. Once the circulating production phase begins, the surface controller generally executes the closing operation according to the wellbore dynamic relationship corresponding to the target working valve. In actual operation, the liquid drop from the previous cycle will change the local tubing pressure at the start of the next cycle. For example, after the previous cycle is drained, some liquid falls back along the tubing wall and reaches the vicinity of the shallow unloading valve, causing the tubing pressure at that location to rise briefly. When the pressure is repressurized in the next cycle, the deep working valve may remain at a small opening due to insufficient pressure difference on both sides, and the shallow unloading valve that has already stopped gas injection may reopen. At this time, the actual inlet position of the high-pressure gas moves upward, while the casing pressure and total gas injection flow rate collected from the ground may still be similar to those when gas is injected through the deep working valve.

[0003] Existing circulating gas lift control typically adjusts pressurization parameters based on the deviation between surface feedback and setpoint, and uses a pre-defined wellbore model to describe the correspondence between control commands and wellhead responses. Some gas lift valve detection methods can determine the actual gas passage location using tracer media, downhole acoustic signals, or dedicated monitoring devices, providing a basis for downhole equipment status analysis. In gas lift truck operations without continuous downhole valve position feedback, the surface control system still needs to use pressure and flow responses for real-time control. When the shallow unloading valve and the deep working valve are in a critical competition state, adjustments to the target pressure or pressurization rate will change the gas injection volume. This could also change the actual downhole channel involved in gas injection. If the controller continues to call the fixed wellbore model and writes the feedback generated by different gas injection channels into the same set of adaptive parameters, the same surface control command will correspond to different pressure propagation paths and liquid carrying processes, causing the control model to absorb conflicting control gains at the same time. As a result, the boosting parameters may be repeatedly adjusted between multiple value ranges, and subsequent control actions may cross the opening boundary of the gas lift valve again. Therefore, the industrial control of circulating gas lift needs to further solve the problems of control topology identification, model selection, and parameter update access under the condition that the downhole gas injection channel is not directly measurable. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: An adaptive control method for circulating air lift boosting parameters includes: Obtain the operating data of the circulating gas lift well, establish a set of candidate gas injection topologies according to the wellbore structure and gas lift valve conversion relationship, use the state model corresponding to each candidate gas injection topology to constrain the operating data, and form the current state set of each candidate gas injection topology. When each current state set is within the reference recovery state set, the control boosting actuator executes the forward boosting trajectory and the reverse boosting trajectory with opposite boosting order, forming a reverse boosting pairing record; The surface metering cumulative mass is corrected based on the changes in the high-pressure gas pipeline inventory mass. The effective wellbore cumulative mass range is then aligned with the forward-sequence pressurization trajectory and the reverse-sequence pressurization trajectory to generate a reverse-sequence differential response set. The inverse differential response set is subjected to continuous quality interval verification with the inverse differential reachability domain of each candidate gas injection topology to form a topology identification record; The topology transformation parameter set is updated based on conditional topology transformation evidence, the control parameter set within the channel is updated based on the fixed topology response for which no conditional topology transformation evidence has been formed, and the topology identification record without shared test cycles is used for verification to form a topology control certificate. Read the topology control certificate and each current state set, propagate the candidate annular pressure reference sequence of each candidate gas injection topology in parallel, filter the common annular pressure reference sequence according to the control constraints, execute the first reference value of the common annular pressure reference sequence, and re-perform state constraints and sequence filtering based on the operating data of the next control cycle; If no common annular pressure reference sequence is formed, and all current state sets are within the recovery access state set, then fixed recovery control is executed and parameter updates are frozen. If any current state set is outside the corresponding recovery access state set, then the field fixed safety control procedure is executed.

[0005] Furthermore, the candidate gas injection topology set is established according to the gas lift valve installation order and permissible conversion relationships; The state constraints include: propagating the corresponding state set using the state model corresponding to each candidate gas injection topology, intersecting the propagation result with the pressure measurement constraints to form the current state set of each candidate gas injection topology, and removing candidate gas injection topologies with empty intersections from the candidate gas injection topology set.

[0006] Furthermore, the reference recovery state set is formed by the state envelope of multiple stable operating cycles under fixed recovery control; Calculate the bidirectional coverage between the current state sets of corresponding candidate gas injection topologies in two running cycles. When the candidate gas injection topology sets of the two running cycles are consistent and the bidirectional coverage satisfies the pairing condition, execute the forward-order boosting trajectory and the reverse-order boosting trajectory respectively. The forward and reverse pressurization trajectories use the same total duration and ground-based metering mass budget, and execute high and low pressurization in reverse order.

[0007] Furthermore, the formation of the effective cumulative mass range for wellbore penetration includes: The inventory mass range is calculated based on the pressure, temperature, and effective volume of the gas control body between the flowmeter measurement section and the effective injection boundary at the wellhead. The inventory mass change is generated based on the inventory mass range at the start time of the identification window and the current time. The inventory mass change is used to correct the cumulative mass measured on the ground, and a non-reduction path verification is performed on the effective cumulative mass range injected into the well.

[0008] Furthermore, the formation of topological identification records includes: A common mass node is set in the overlapping interval of the effective cumulative mass range of the positive-sequence pressurization trajectory and the negative-sequence pressurization trajectory, and a negative-sequence differential response set is formed on the common mass node; Using common wellbore structural parameters and gas lift valve parameters, the forward-order boosting trajectory and the reverse-order boosting trajectory are jointly propagated according to the state model corresponding to each candidate gas injection topology, forming the reverse-order differential reachable domain of each candidate gas injection topology; Compare the inverse differential response set with the corresponding inverse differential reachable domain along the continuous mass interval. Remove candidate gas injection topologies from the candidate gas injection topology set whose continuous inconsistent mass span reaches the continuous exclusion mass length. Write the retained candidate gas injection topologies and conditional topology transformation evidence into the topology identification record.

[0009] Furthermore, updates to the topology transformation parameter set include: Maintain a set of topology transformation parameters for each air lift valve, state before transformation, state after transformation, and direction of transformation. Based on the conditional topology transformation evidence in two topology identification records that do not share test cycles, form valve depth differential pressure evidence for the same air lift valve state transformation. The valve depth differential pressure evidence is not trimmed using the set of topology transformation parameters to be updated. Find the intersection of the two valve depth differential pressure evidences with the topology transformation parameter set adjusted according to the parameter drift, and write the intersection result into the topology control certificate.

[0010] Furthermore, the update of the control parameter set within the channel includes: Select topology identification records that retain only one candidate gas injection topology and do not form conditional topology transformation evidence, and perform constraint intersection on the set of control parameters in the channel based on the fixed topology response; The parameter direction separation degree is calculated based on the difference in wellhead pressure response corresponding to different parameter directions. When the parameter direction separation degree meets the admission conditions, a set of control parameters in the channel to be verified is formed. Verification is performed using a topology identification record that does not share the test cycle with the topology identification record that forms the control parameter set within the channel to be verified, and the verified control parameter set within the channel is written into the topology control certificate.

[0011] Further screening of the common annular pressure reference series includes: The applicable topology scope of the verification topology control certificate includes the candidate gas injection topology set. The candidate annular pressure reference sequence is propagated according to each candidate gas injection topology and its allowed switching branches. The candidate annular pressure reference sequence that cannot be executed is screened out based on the equipment operating boundary, gas lift valve switching boundary and terminal recovery condition. Candidate annular pressure reference sequences are selected in order of increasing wellhead pressure prediction deviation and decreasing gas lift valve switching boundary remaining interval to form a common annular pressure reference sequence.

[0012] Furthermore, when executing the common annular pressure reference sequence, only the first reference value is issued to the booster actuator, and the state constraints and sequence selection are re-performed in the next control cycle; When no common annular pressure reference sequence is formed, the current state set of each candidate gas injection topology is compared with the corresponding recovery access state set. When each current state set is within the corresponding recovery access state set, fixed recovery control is executed, and the updates of the topology transformation parameter set and the channel control parameter set are frozen. When any current state set is outside the corresponding recovery access state set, the on-site fixed safety control procedure is executed.

[0013] An adaptive control system for circulating air lift boosting parameters includes: The state observation module establishes a set of candidate gas injection topologies based on the wellbore structure and the gas lift valve conversion relationship. It uses the state model corresponding to each candidate gas injection topology to constrain the operation data of the circulating gas lift well, forming the current state set of each candidate gas injection topology. The reverse pressurization identification module controls the pressurization actuator to execute the forward pressurization trajectory and the reverse pressurization trajectory when all current state sets are within the reference recovery state set. It corrects the cumulative mass of surface metering based on the changes in the high-pressure gas pipeline inventory, aligns the forward pressurization trajectory and the reverse pressurization trajectory with the effective cumulative mass range of the wellbore, and performs continuous mass interval verification between the reverse differential response set and the reverse differential reachable domain of each candidate gas injection topology to form a topology identification record. The parameter update module updates the topology transformation parameter set based on the conditional topology transformation evidence in the topology identification record, updates the control parameter set within the channel based on the fixed topology response that does not form conditional topology transformation evidence, and verifies the updated parameter set using the topology identification record that does not share test cycles to form a topology control certificate. The rolling control module reads the topology control certificate and each current state set, propagates the candidate annular pressure reference sequence of each candidate gas injection topology in parallel, filters the common annular pressure reference sequence according to control constraints, executes the first reference value of the common annular pressure reference sequence, and re-performs state constraints and sequence filtering based on the operating data of the next control cycle.

[0014] This invention provides an adaptive control method and system for circulating air lift boosting parameters, which has the following beneficial effects: 1. This invention constructs path differences under the same mass budget using forward-order and reverse-order pressurization trajectories with opposite pressurization sequences. Through high-pressure gas path inventory mass compensation, the surface metered gas injection volume is converted into effective wellbore cumulative mass coordinates, making the pressure propagation differences corresponding to shallow unloading valve gas passage, valve switching, and deep working valve gas passage apparent from the surface pressure and flow response. Thus, even without continuous downhole valve position feedback, it is still possible to eliminate inconsistent candidate gas injection topologies based on the reverse-order differential response within the continuous mass range, avoiding misjudging operating states with similar surface responses but different actual gas injection locations as the same control state.

[0015] 2. This invention incorporates the topology transformation process and the fixed gas injection channel process into the parameter update chain: the topology transformation parameters are constrained by the recurring conditional topology transformation evidence, and the control parameters within the channel are constrained by the response under a unique fixed topology, and the update verification is completed through independent operation records; this process cuts off the parameter sharing relationship between different downhole gas injection channels, prevents the shallow unloading valve response and the deep working valve response from being written into the same control gain, and ensures that the adaptive parameters are always bound to the corresponding gas injection topology, suppressing the repeated adjustment of the boosting parameters between multiple value ranges.

[0016] 3. In the online control phase, this invention propagates the annular pressure reference sequence under the current candidate gas injection topology in parallel, and only executes common control commands that satisfy the equipment boundary, gas lift valve switching boundary, and terminal recovery conditions for each candidate gas injection topology. When the downhole gas injection channel is not yet uniquely determined, the controller does not need to pre-select a single wellbore model. When there is no common executable sequence, it switches to a fixed recovery control or fixed safety control program, thereby avoiding the control action from crossing the unconfirmed gas lift valve switching boundary again, so that topology identification, parameter update, and pressurization execution form a feedback update control chain constrained by the admission conditions. Attached Figure Description

[0017] Figure 1 This is a schematic diagram showing the relationship between the candidate gas injection topology of the circulating gas lift well and the ground industrial control system of the present invention. Figure 2 This is a schematic diagram illustrating the high-pressure gas path inventory quality compensation and effective well-entry cumulative quality formation of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 and Figure 2 This embodiment provides an adaptive control method for circulating air lift boosting parameters, including the following specific steps: The following is an explanation of the continuous control process of a circulating gas lift well in a gas field from the 37th to the 66th operating cycles. After the well was drained in the previous operating cycle, some of the liquid fell back to the vicinity of the second unloading valve. When pressurizing in the next operating cycle, the surface annulus pressure and the surface gas injection mass flow rate were still within the normal range, but the second unloading valve may re-enter the gas flow, and the working valve may also open with a delay, so that the wellhead pressure response obtained by using the fixed wellbore model is inconsistent with the actual response.

[0020] This embodiment uses the same well, the same set of booster devices, the same set of sensors, and the same version of the wellbore model as the processing objects, and sequentially forms a reverse booster pairing record P1, a topology identification record P2, a topology control certificate C3, and a control execution record P4. Unless otherwise stated, all pressures are absolute pressures, and all surface gas injection flow rates are mass flow rates after temperature and pressure compensation.

[0021] This embodiment is applied to a circulating gas lift well that uses casing annulus gas injection and tubing fluid drainage. Two unloading valves and one working valve are installed downhole along the depth direction. The installation depths of the first unloading valve, the second unloading valve, and the working valve are 780m, 1450m, and 2180m, respectively. The design data of the three gas lift valves provide the opening pressure differential range, the closing pressure differential range, and the permissible state transition sequence. The surface equipment includes a circulating gas lift booster device, a wellhead gas injection valve, an annulus pressure transmitter, a wellhead pressure transmitter, a temperature and pressure compensated mass flow meter, and pressure and temperature measuring points in the high-pressure gas path. The booster controller uses the annulus pressure reference value as the outer loop control command, and the inner pressure loop generates compressor load commands and wellhead gas injection valve opening commands.

[0022] In this embodiment, the pressure range, mass range, time range, and control command range are all stored in the form of closed intervals. The state set, model parameter set, and feasible control set are stored using interval upper and lower bounds or a finite number of linear inequalities. When two intervals intersect, the larger of the two lower bounds is taken as the lower bound of the intersection, and the smaller of the two upper bounds is taken as the upper bound of the intersection. When the lower bound of the intersection is greater than the upper bound of the intersection, the intersection is considered empty. When the parameter set involves the correlation constraints between multiple parameters, the linear inequalities that form the parameter set are saved, and the original parameter set is not replaced by the independent projection range of each parameter. The individual parameter ranges listed below are the projection ranges of the complete parameter set in the corresponding parameter direction.

[0023] The 34th to 36th operating cycles used a fixed working valve model for pressurization regulation. During the 35th operating cycle, when the annular pressure increased from 9 MPa to 13.5 MPa, the wellhead pressure lagged behind the model prediction by about 9 seconds. During the 36th operating cycle, under the condition that the surface gas injection quality and the terminal annular pressure were close, the wellhead pressure advanced by about 6 seconds compared to the model prediction. Since there was no continuous downhole valve position sensor on site, the state transition of the downhole gas lift valve could not be determined solely based on the advance or lag of the wellhead pressure in a single cycle.

[0024] Data on annular pressure, wellhead pressure, surface gas injection mass flow rate, and high-pressure gas path status are collected using annular pressure transmitters, wellhead pressure transmitters, temperature-compensated mass flow meters, and high-pressure gas path pressure and temperature measuring points. The actual compressor load, actual wellhead injection valve opening, compressor load command, and wellhead injection valve opening command are read from the booster controller. Simultaneously, the currently effective injected gas component detection records and their version identifiers are read. When the production control system can provide external backpressure with a unified timestamp, the external backpressure is used only as a check factor for the model's applicability. When no valid external backpressure data is provided, the external backpressure is not used as the input for determining the candidate gas injection topology; its impact is absorbed by the process disturbance range.

[0025] Each data channel uses a unified clock to record the sampling timestamp. The fixed processing delay of the sensor and acquisition channel is obtained by applying a step signal to the input of the corresponding transmitter through a signal calibrator during the commissioning calibration phase. The preferred range for the number of repeated calibrations is 5 to 8 times. In this embodiment, 6 times are used. Each calibration records the step signal input time and the start time when the controller received the value for 3 consecutive unified control cycles and entered the range of the final step change amount ±5%. The difference between the two times forms the delay interval. The intersection of the 6 effective delay intervals is calculated, and the center value of the intersection is used as the fixed processing delay of the channel. The half-width of the intersection is included in the time synchronization error. If there is no common intersection among the delay intervals, or if the delay change range exceeds the time synchronization tolerance, the fixed processing delay calibration of the channel is invalid, and the current data cannot be used for active identification. The fixed processing delay of the sensor and acquisition link is calibrated and recorded separately from the high-pressure gas path propagation delay and the wellbore pressure propagation delay.

[0026] In this embodiment, the fixed processing delay for each channel is 0.6s to 1s, and the preferred range for the unified control cycle is 0.2s to 1s. In this embodiment, 0.5s is used. This value is not greater than 1 / 5 of the maximum action time of the wellhead gas injection valve (3s). It can obtain no less than 5 sets of feedback data before the actuator completes a significant action. The preferred range for the time synchronization tolerance is 0.05s to 0.2s. In this embodiment, 0.1s is used. At the maximum stable pressure rise slope of 0.16MPa / s, the pressure misalignment caused by the 0.1s synchronization error does not exceed 0.016MPa, which is less than the upper limit of the comprehensive measurement error of the pressure transmitter (0.03MPa). When the same control moment data group cannot be formed within the time synchronization tolerance, the current data group does not participate in state observation, mass integration, and trajectory validity confirmation.

[0027] The high-pressure gas path from the compressor outlet to the effective injection boundary at the wellhead is divided into three sections: an outlet rigid pipe, a buffer container, and a high-pressure hose, each with a calibrated effective volume of 0.012 m³. 3 0.026m 3 and 0.009m 3 After considering the change in elastic volume under working pressure, the effective volume range of each segment is taken as 97% to 103% of the corresponding calibration value. The range is determined by the geometric volume measurement results under normal pressure and the volume change test under 16.0MPa pressure.

[0028] The pressure transmitter has a range of 0 MPa to 20 MPa, with an upper limit of comprehensive measurement error of 0.03 MPa; the mass flow rate has a range of 0 kg / s to 0.8 kg / s, with a maximum permissible error of 0.005 kg / s. The instrument calibration results, gas path volume calibration results, injected gas component detection results, and pressure step response test results are written into the corresponding version records. When the sensor, high-pressure hose, buffer container, injected gas component, or gas physical property parameters change, a new version identifier is activated, and different version data are not combined to form the same identification record.

[0029] Based on the number of downhole gas lift valves, installation depth, design opening differential pressure, design closing differential pressure, and allowable state transition order, a set of candidate gas injection topologies is established. In this embodiment, four candidate gas injection topologies are set: "Continuous gas flow from the first unloading valve", "Transition from the first unloading valve to the second unloading valve", "Transition from the second unloading valve to the working valve", and "Continuous gas flow from the working valve". The candidate gas injection topology represents the gas injection channel state at the beginning of an identification window and the state transitions allowed to occur within that identification window. Transitions not supported by the wellbore design data are not included in the set of candidate gas injection topologies. A bounded state observer is established for each candidate gas injection topology. The state quantities include annular pressure, wellhead pressure, annular pressure decay rate, and wellhead pressure recovery residual.

[0030] The annular pressure decay rate is formed using the pressure sequence from 8s to 12s at the end of the recovery phase. In this embodiment, a 10s time window is used. During the calculation, each annular pressure sample value within the time window is extended into a pressure interval according to the upper bound of the measurement error. For any two sampling moments with a time interval of not less than 2s, the pressure interval of the second moment is subtracted from the pressure interval of the first moment, and then divided by the time interval between the two moments to obtain the corresponding slope interval. The minimum value among the lower bounds of all slope intervals and the maximum value among the upper bounds of all slope intervals are taken to form the annular pressure decay rate range.

[0031] The reference recovery pressure trajectory for each candidate gas injection topology is formed by stable cycles that have been confirmed during the commissioning and calibration phase, using the same fixed recovery control program and the same model version. Each stable cycle is aligned according to the relative time after the start of recovery control, and the envelope of the wellhead pressure range at the same relative moment is calculated. After the number of stable cycles and the envelope convergence condition are met, the trajectory is frozen as the reference recovery pressure trajectory for the corresponding candidate gas injection topology. The wellhead pressure recovery residual is the interval difference between the measured wellhead pressure range and the corresponding reference recovery pressure trajectory range under the same recovery duration, with a pressure filtering delay of 10s time window longer than 1s, and covering the 7s surface high-pressure gas path propagation delay.

[0032] The applicable range of annular pressure is segmented according to pressure intervals from 0.8MPa to 1.2MPa. In this embodiment, it is segmented according to 1MPa. Within each pressure segment, a set of state transition parameters and a set of control action parameters are formed by pressure step tests of the gas injection topology confirmed during the commissioning phase. Within each pressure segment, the state model of the candidate gas injection topology uses the current state range, the current control input range, the set of state transition parameters, the set of control action parameters, and the range of process disturbances as inputs. When calculating the state at the next control moment, the current state range is first propagated using the set of state transition parameters. Then, the control actions generated by the change of annular pressure reference value, the compressor load range, and the wellhead gas injection valve opening range are superimposed according to the set of control action parameters. Finally, the range of process disturbances is superimposed to form the predicted state set for the next control moment.

[0033] For a candidate gas injection topology that includes gas lift valve state transition, the state transition parameter set and control action parameter set corresponding to the gas injection channel before the state transition are used. When the valve depth differential pressure prediction envelope overlaps with the corresponding topology transition parameter set, the prediction branch of the gas lift valve state transition is retained, and the parameter set corresponding to the gas injection channel after the transition is used to continue propagation. When the valve depth differential pressure prediction envelope crosses the transition boundary but cannot uniquely confirm that the gas lift valve has transitioned, both the branch before and the branch after the transition are retained, and the two branches are not merged into a single state set.

[0034] The preferred range for the number of independent effective step tests for each pressure segment is 6 to 10 sets. In this embodiment, 8 sets are used. The tests cover the pressure increase and decrease directions, as well as the high pressure increase slope, low pressure increase slope, main compressor load and wellhead gas injection valve opening range used in subsequent identification. If there are fewer than 6 effective step tests for a certain pressure segment, an online identification model for that pressure segment will not be formed.

[0035] For each set of step tests, the current state range, control input range, and measured state range at the next control moment are substituted into the piecewise state model to form linear constraints on the state transition parameters and control action parameters. All parameter combinations that can make the one-step predicted state intersect with the measured state range are retained. The feasible parameter sets formed by the 8 sets of step tests in the same pressure segment are intersected to obtain the set of state transition parameters and the set of control action parameters for that pressure segment.

[0036] When the feasible set of parameters is empty, it means that the pressure segment cannot be explained by the current segment state model. The model is not formed by relaxing the parameter range or deleting unfavorable test records. After the parameter set is formed, it is verified by two independent stable cycles that do not participate in the parameter calculation. If the measured state does not intersect with the predicted state set in any of the five consecutive unified control cycles in any verification cycle, the pressure segment model shall not be used for online identification.

[0037] For each stable cycle with a confirmed gas injection topology, the interval difference between the measured state range at the next moment and the state model prediction range is calculated. The envelope of the residual range of all effective stable cycles within the same pressure segment is obtained to form the basic process disturbance range. The changes in conventional external back pressure and wellbore fluid replenishment during the commissioning calibration period are covered by the stable cycle residual envelope. During online operation, the two are no longer fitted separately as new gas lift valve state transitions.

[0038] Changes in external back pressure and wellbore fluid replenishment are treated as continuous state disturbances, while the opening or closing of the gas lift valve is treated as a discrete state transition in the candidate gas injection topology. External back pressure data with valid timestamps are only used to determine whether the current operating condition is within the model calibration range. When external back pressure data is not connected, the estimated external back pressure value is not used as the state model input. When the measured state exceeds the bounded disturbance range, the current data is marked as insufficient for the model's applicability range, and external operating conditions exceeding the range are not explained by adding gas lift valve state transitions.

[0039] The fixed recovery control program aims to control the decrease in annular pressure. It limits the compressor load to 20% to 35% of the rated load, the wellhead injection valve opening to 10% to 25%, and the annular pressure decrease rate to 0.02 MPa / s to 0.04 MPa / s. The lower limits of the compressor load and wellhead injection valve opening are determined by the equipment's minimum stable operating load and the valve's stable adjustment opening, respectively. The upper limits are determined by the maximum command in the fixed recovery test that does not trigger the unconfirmed gas lift valve switching and can maintain controlled pressure reduction. In this embodiment, a 30% compressor load and an 18% wellhead injection valve opening are used as the initial recovery command, with 0.03 MPa / s as the preferred decrease rate. During execution, the inner pressure loop corrects the command within the range. If the decrease rate cannot be maintained, the program switches to the field fixed safety control program.

[0040] The on-site fixed safety control program is pre-installed in the equipment controller. Its actions include canceling the active pressure boost reference value, adjusting the compressor load to the equipment's minimum stable operating load, adjusting the wellhead gas injection valve opening according to the allowable rate of change of the wellhead gas injection valve's action calibration, and maintaining the highest execution priority for hardware interlocks of equipment overpressure, compressor overload, and valve failure.

[0041] During the commissioning calibration phase, the fixed recovery control program is continuously executed and the state set at the end of the recovery is collected. The preferred range for the number of independent stable cycles used to form the reference recovery state set is 5 to 12, and 8 are used in this embodiment. The preferred range for the number of continuous confirmation cycles of the recovery state envelope is 2 to 4, and 3 are used in this embodiment. When the effective independent stable cycles are less than 5, the reference recovery state set is not frozen and the reverse-order boost active identification is not initiated.

[0042] If the newly added stable cycle does not cause the envelope boundary of any state direction to expand beyond the sum of the corresponding measurement error bound and the model residual bound, and this condition remains true for 3 consecutive newly added cycles, the current envelope is frozen as the reference recovery state set. If the condition is still not met after 12 collection cycles, the active identification and verification of the state model is stopped, and the reference recovery state set is not further expanded.

[0043] The reference recovery state set after freezing in this embodiment includes: annular pressure of 8.6 MPa to 9 MPa, wellhead pressure of 3.1 MPa to 3.4 MPa, annular pressure decay rate with an absolute value not greater than 0.004 MPa / s, and wellhead pressure recovery residual with an absolute value not greater than 0.1 MPa. The reference recovery state set remains read-only during online operation. In each running cycle, pressure, flow rate, execution feedback, and control commands are read according to a unified clock. After completing the fixed processing delay correction for each channel, the pressure data forms measurement constraints. The surface gas injection mass flow rate and execution feedback are used for mass integration and trajectory validity confirmation. Control commands are used for state set propagation. For each candidate gas injection topology, the current state set is substituted into the corresponding segmented state model, and interval propagation is performed on the state transition parameter set, control action parameter set, and bounded disturbance range to form the predicted state set for the next control moment. Then, the predicted state set is intersected with the pressure measurement constraints to form the updated state set. The state sets of different candidate gas injection topologies are saved separately, and the state sets of different candidate gas injection topologies are not merged or convex hulls are calculated.

[0044] When the state set of a candidate gas injection topology is empty, the candidate gas injection topology is excluded from the current state observation window. When the state sets of all candidate gas injection topologies are empty, active identification stops and the on-site fixed safety control procedure is executed. The exclusion of a candidate gas injection topology in the identification window that forms a certain P2 only indicates that the candidate gas injection topology cannot explain the corresponding reverse-order pressurization test pair, and does not constitute a permanent deletion for subsequent operating cycles. At the start of a new operating cycle, the current candidate gas injection topology set is still formed based on the candidate gas injection topologies allowed by the wellbore structure and the current state observation results. After the routine fluid drainage is completed, the fixed recovery control procedure is executed. The allowable expansion range of the reference recovery state set is calculated by summing the measurement error boundaries of each state direction and the maximum stable residual of the fixed recovery test. The frozen reference recovery state set is expanded item by item. In this embodiment, the expansion amounts of annular pressure, wellhead pressure, annular pressure decay rate, and wellhead pressure recovery residual direction are 0.2MPa, 0.12MPa, 0.004MPa / s, and 0.1MPa, respectively, and written to the current configuration version. When the current state set corresponding to each candidate gas injection topology is within the allowable expansion range and the recovery confirmation time is continuously reached, the current running cycle meets the recovery admission conditions.

[0045] The preferred range for the recovery confirmation time is 45s to 120s, and its lower limit is not less than the sum of the longest state stabilization time and the synchronization margin obtained from the fixed recovery test. In this embodiment, the longest state stabilization time is 52s, and the synchronization and execution margin is 8s. Therefore, the recovery confirmation time is 60s.

[0046] The first operating cycle that meets the recovery access conditions is designated as the first test cycle. After the first test cycle is completed, at least one regular operating cycle that does not participate in active identification is executed, and the fixed recovery control program is executed again. The preferred range for the number of regular operating cycles is 1 to 3. In this embodiment, 1 cycle is executed first. If the initial state still does not meet the pairing conditions, the number is increased to 3. If pairing still cannot be achieved after 3 cycles, the active identification is canceled. The state sets corresponding to each candidate gas injection topology at the beginning of the two test cycles are saved respectively. The bidirectional coverage is calculated based on the allowable deviation of annular pressure, wellhead pressure, annular pressure decay rate, and wellhead pressure recovery residual. When calculating the normalized distance between states, Calculate the minimum interval between the two states in the directions of annular pressure, wellhead pressure, annular pressure decay rate, and wellhead pressure recovery residual. Then divide each minimum interval by the corresponding allowable deviation and take the maximum value of the four ratios as the normalized distance between the two states. For each state in the first test cycle state set, calculate its minimum normalized distance to the second test cycle state set and take the maximum value of the minimum normalized distance. Then swap the two state sets and repeat the calculation. Take the larger of the two results as the bidirectional coverage. The bidirectional coverage is not greater than 1. When the candidate gas injection topology sets at the beginning of the two test cycles are the same, it is allowed to form a reverse-order pressurization test pair.

[0047] The preferred range for the permissible deviation of annular pressure is 0.15 MPa to 0.25 MPa, and 0.2 MPa is used in this embodiment; the preferred range for the permissible deviation of wellhead pressure is 0.1 MPa to 0.15 MPa, and 0.12 MPa is used in this embodiment; the preferred range for the permissible deviation of annular pressure decay rate is 0.003 MPa / s to 0.005 MPa / s, and 0.004 MPa / s is used in this embodiment; the preferred range for the permissible deviation of wellhead pressure recovery residual is 0.08 MPa to 0.12 MPa, and 0.1 MPa is used in this embodiment. Each preferred value is determined by the sum of the upper limit of the corresponding instrument error and the maximum stable residual of the fixed recovery test, and is not relaxed during online operation due to pairing requirements.

[0048] The 37th running cycle was selected as the first test cycle. At the beginning of its identification window, the annular pressure was 8.82 MPa, the wellhead pressure was 3.26 MPa, the annular pressure decay rate was -0.0027 MPa / s, and the wellhead pressure recovery residual was 0.04 MPa. After the 37th running cycle was completed, one regular running cycle and a fixed recovery control program were executed. The 40th running cycle was selected as the second test cycle. Its corresponding four state variables were 8.86 MPa, 3.23 MPa, -0.003 MPa / s, and 0.03 MPa, respectively. The normalized bidirectional coverage of the two test cycles in each state direction was 0.72. At the beginning of both test cycles, four candidate gas injection topologies were retained: "continuous gas flow from the first unloading valve", "transition from the first unloading valve to the second unloading valve", "transition from the second unloading valve to the working valve", and "continuous gas flow from the working valve", which satisfied the pairing condition of the initial state and the candidate gas injection topology set.

[0049] A single identification window starts from the control moment when the annular pressure reference trajectory of the first boosting stage is received from the pressure inner loop and ends at the control moment when the predetermined duration of the second boosting stage ends. If an interlock or invalid condition occurs during trajectory execution, the identification window ends at the moment the termination control command is issued. The recovery process after termination is not included in the identification window. After pairing is established, the forward-sequence boosting trajectory and the reverse-sequence boosting trajectory are solved. The forward-sequence boosting trajectory executes the high boosting slope and the low boosting slope in sequence, and the reverse-sequence boosting trajectory executes the low boosting slope and the high boosting slope in sequence. Both trajectories are used as annular pressure reference trajectories and input into the same version of the boosting control inner loop. The fixed pressure tracking test is conducted within the range of compressor load not exceeding 90% and wellhead gas injection valve opening of 10% to 85%, obtaining a stable tracking slope range of 0.02MPa / s to 0.16MPa / s, with an upper limit of slope error of 0.015MPa / s.

[0050] The preferred range for a low pressure rise slope is 0.04 MPa / s to 0.07 MPa / s, and in this embodiment, it is 0.05 MPa / s. The preferred range for a high pressure rise slope is 0.09 MPa / s to 0.13 MPa / s, and in this embodiment, it is 0.11 MPa / s. The difference between the high and low pressure rise slopes is 0.06 MPa / s, which is greater than twice the upper limit of the slope error of 0.03 MPa / s. The time from the annular pressure reference value to the formation of a stable gas injection response at the effective injection boundary at the wellhead includes the compressor response time, the wellhead gas injection valve action time, the high-pressure gas path propagation delay, and the pressure acquisition stabilization time. In this embodiment, the response times are approximately 5s, 3s, 7s, and 1s, respectively, with a total time of approximately 16s.

[0051] The preferred range for the duration of a single boost phase is 20s to 40s, with a lower limit of at least 4s of stable response observation time remaining outside the total response time of 16s. In this embodiment, the duration of the high boost phase is 25s, and the duration of the low boost phase is 35s. The preferred range for the total duration of the two trajectories is 50s to 70s, and in this embodiment, it is 60s. The reverse boost phase only swaps the execution order of the two boost phases without changing the slope, duration, or total duration of the two boost phases. The preferred range for the common ground-based metrological mass budget is 15kg to 30kg, with the lower limit determined by the minimum identified mass of 12kg superimposed on the mass mapping and the model uncertainty margin, and the upper limit determined by the hard constraints of annular pressure of 16MPa and compressor load of 90%. In this embodiment, it is 22kg.

[0052] The preferred range for mass pairing error is 0.7 kg to 1 kg. Calculated as 0.76 kg based on the sum of the mass corresponding to twice the flowmeter integral error and the time synchronization error, this embodiment takes 0.8 kg upwards. For each feasible combination of pressurization slope and duration, the pressure response range of each candidate gas injection topology under the positive and negative pressurization trajectories is predicted. Along the three response directions of annular pressure, wellhead pressure, and the difference between annular pressure and wellhead pressure, the non-overlapping distance between the predicted response ranges of any two candidate gas injection topologies is calculated and divided by the measurement error and model residual envelope in the corresponding direction.

[0053] The minimum value of all candidate gas injection topology combinations is taken to form the minimum topological separation degree of the current trajectory parameters, with a theoretical lower limit of 1. Considering sampling discrepancy and model set out-of-body error, the preferred range of the active identification admission threshold is 1.1 to 1.3, and 1.2 is taken in this embodiment. The preferred range of the minimum effective non-overlapping amount in the pressure direction is 0.22MPa to 0.30MPa. In this embodiment, the combined upper limit of the pressure measurement error, model residual, time synchronization error and mass mapping set out-of-body error of the two trajectories is 0.21MPa, and is taken upward to 0.25MPa.

[0054] Equipment pressure, compressor load, wellhead injection valve action, unconfirmed gas lift valve switching, and pressure inner loop tracking capability are used as hard constraints. After the selected trajectory simultaneously satisfies the minimum topology separation degree of not less than 1.2 and the minimum effective non-overlap in the pressure direction of not less than 0.25MPa, the time average value of the upper limit of the compressor load corresponding to all candidate injection topologies is minimized first. When the difference between the time average values ​​of two trajectory combinations does not exceed 1 percentage point, the sum of the absolute values ​​of the changes in the annular pressure reference slope in adjacent control cycles is minimized, and the obtained trajectory combination is selected.

[0055] The first test cycle executes the forward-sequence pressurization trajectory, and the second test cycle executes the reverse-sequence pressurization trajectory. The preferred range for the actuator tracking anomaly confirmation time is 3s to 8s, and 5s is used in this embodiment. This value is not shorter than the 3s action time of the wellhead gas injection valve. During trajectory execution, if the annular pressure reaches 16.0MPa, the compressor load reaches 90%, the error between the actual opening and the commanded opening of the wellhead gas injection valve lasts for more than 5s, any pressure channel signal is interrupted, the surface gas injection mass flow rate signal is interrupted, the unified time synchronization fails, or the equipment hardware interlock is activated, the current trajectory is immediately terminated and the parameter update is frozen. If the hardware interlock is not latched and all current state sets are within the pre-formed offline recovery access state set, the fixed recovery control program is executed. If the hardware interlock is latched or there is a current state set outside the corresponding recovery access state set, the on-site fixed safety control program is executed or the equipment hardware interlock takes over. The terminated trajectory must not form a reverse-sequence pressurization test pair with another trajectory.

[0056] The 37th running cycle's positive-sequence boosting trajectory lasted 60.1 s, with a cumulative surface metering mass of 22.5 kg, a terminal annular pressure of 13.68 MPa, and a terminal wellhead pressure of 4.05 MPa. The 40th running cycle's negative-sequence boosting trajectory lasted 59.8 s, with a cumulative surface metering mass of 22.3 kg, a terminal annular pressure of 13.51 MPa, and a terminal wellhead pressure of 3.88 MPa. The difference in duration between the two trajectories was 0.3 s, and the difference in cumulative surface metering mass was 0.2 kg, both meeting the pairing requirements. Within the annular pressure range of 11 MPa to 12 MPa, the wellhead pressure rise of the positive-sequence boosting trajectory preceded that of the negative-sequence boosting trajectory, forming a path-related response for subsequent topology identification processing.

[0057] After both trajectories are successfully completed, a reverse-order pressurization pairing record P1 is generated. P1 uses the well identifier, the first test cycle identifier, the second test cycle identifier, the unified clock identification window, and the configuration version identifier as association keys. It also saves the initial state set, forward-order pressurization trajectory, reverse-order pressurization trajectory, trajectory parameters, original annular pressure, wellhead pressure, surface gas injection mass flow rate, high-pressure gas path pressure and temperature sequence, injected gas component detection record, actual execution record, terminal annular pressure, terminal wellhead pressure, sampling timestamp, flow meter installation location, effective volume range of each high-pressure gas path segment, and all model and configuration version identifiers. Data that cannot be uniquely mapped to the same well and the same test pair by the association key is not written to the same P1.

[0058] After P1 is formed, all version identifiers are checked. The gas control body between the flowmeter measurement section and the effective injection boundary at the wellhead should maintain a single inlet and single outlet state within the identification window. When unloading, venting, bypassing, branch switching, or reverse flow occurs within the identification window, the current P1 fails. The absolute pressure applicable range for the three high-pressure gas path segments is 8MPa to 16MPa, and the absolute temperature applicable range is 285K to 325K. The gas compressibility factor is obtained by inputting the same pressure, temperature, and injected gas composition state into the corresponding version of the physical property model. If the pressure, temperature, or gas composition exceeds the calibration range of the physical property model, or if any segment lacks effective pressure or temperature data, the inventory mass calculation is stopped, and the current P1 does not enter topology identification.

[0059] For each high-pressure gas path segment, the absolute pressure and effective volume at the same state point are multiplied, and then divided by the product of the gas compressibility factor, the injected gas specific gas constant, and the absolute temperature at that pressure and temperature to obtain the segment's inventory gas mass at that state point. The minimum and maximum values ​​of the segment's inventory gas mass are calculated within the corresponding pressure range, temperature range, and effective volume range to form the segment's inventory mass range. The inventory mass ranges of the three segments are added together to form the gas control volume inventory mass range. When calculating changes in inventory mass, the inventory mass at the beginning of the identification window and the current time are calculated using the same effective volume error, sensor fixed deviation, and gas property parameters, respectively. Then, the mass difference between the two times is calculated to maintain the correlation of common parameter errors within the same test cycle. The cumulative mass range of surface metering is reduced by the increase in high-pressure gas path inventory to form the effective cumulative mass range of well entry. When the high-pressure gas path inventory decreases, the mass of the released gas is included in the effective cumulative mass of well entry.

[0060] The maximum effective wellbore mass lower bound is recursively calculated up to the current time along the sampling time. When the maximum lower bound exceeds the current effective wellbore mass upper bound, there is no effective wellbore mass path that can fall into each mass range at the same time and maintain non-decreasing quality. The current trajectory is invalid. The recursive result is only used to determine the feasibility of non-decreasing paths and is not used as the only effective wellbore mass curve.

[0061] The effective cumulative mass of the terminal injection into the well should overlap between the positive-sequence pressurization trajectory and the negative-sequence pressurization trajectory. If the effective cumulative mass of the terminal injection into the well does not overlap, the current P1 fails. If there is an overlap, the coverage mass is jointly guaranteed to be the smaller of the lower bounds of the two effective cumulative mass injections into the well. In the 37th operating cycle, the cumulative mass of the surface metering is 22.2 kg to 22.8 kg, the high-pressure gas pipeline inventory increases by 0.7 kg to 1.6 kg, and the effective cumulative mass injection into the well is 20.6 kg to 22.1 kg. In the 40th operating cycle, the corresponding ranges are 22 kg to 22.6 kg, 0.8 kg to 1.8 kg, and 20.2 kg to 21.8 kg, respectively, jointly guaranteeing a coverage mass of 20.2 kg.

[0062] Based on the aforementioned trajectory prediction and separation results, the minimum identification mass in this embodiment is 12 kg, ensuring a coverage mass of 20.2 kg to meet the topology identification requirements. Common mass nodes are established within the range of 0 kg to 20.2 kg. After differential analysis using the same gas path parameters, screening for non-minimum-weighted paths, and time synchronization correction, the effective mass range width of the commonly usable nodes is 1.16 kg to 1.28 kg. Taking the maximum value of the half-width of each node's range as the local mass mapping uncertainty, we obtain a range of 0.58 kg to 0.64 kg. The preferred range for the mass mapping resolution error is 0.55 kg. The mass is between 0.75 kg and 0.65 kg in this embodiment. The preferred range for the common mass grid interval is 0.8 kg to 1.2 kg, while simultaneously satisfying that it is not less than the mass mapping resolution error and not greater than half of the continuous exclusion mass length. In this embodiment, the common mass grid interval is 1 kg. For each common mass node, the corresponding time preimage set is determined within the effective cumulative mass of the two trajectories. When the time preimage set is empty or the span exceeds the allowable mapping duration, the mass node does not participate in the reverse order difference calculation and does not extrapolate data outside the time preimage set.

[0063] The preferred range for the allowed mapping duration is 0.8s to 1.5s, and this embodiment uses 1s. This value is determined based on the larger pressure change rate of 0.11MPa / s within the identification window, ensuring that the pressure range expansion caused by the time preimage span does not exceed 0.11MPa. In this embodiment, the upper limit of the pressure measurement error is 0.03MPa. The upper limit of the outer margin error of the mass coordinate set obtained by replaying the commissioning calibration record using a 1kg common mass grid is 0.04MPa. The sum of the time preimage expansion, pressure measurement error, and mass coordinate set outer margin error does not exceed 0.18MPa. The minimum effective non-overlapping amount of the pressure direction is less than 0.25MPa. For each common mass node, interval subtraction is used: the annular pressure range of the reverse-sequence pressurization trajectory is subtracted from the annular pressure range of the forward-sequence pressurization trajectory to form the annular pressure reverse-sequence difference of that mass node; the wellhead pressure range of the reverse-sequence pressurization trajectory is subtracted from the wellhead pressure range of the forward-sequence pressurization trajectory to form the wellhead pressure reverse-sequence difference of that mass node.

[0064] When forming the cumulative annular-to-wellhead pressure propagation deviation along the common mass coordinate, according to the arrangement order of the common mass nodes, first calculate the interval difference between the annular pressure reverse order difference and the wellhead pressure reverse order difference for each mass node. Then, use the mass interval between two adjacent common mass nodes as the weight of this interval difference. The weighted interval differences between the first common mass node and the current common mass node are accumulated sequentially and normalized according to the cumulative mass span between the first common mass node and the current common mass node to form the pressure propagation deviation range of the current common mass node. When the current common mass node is the first valid node, the interval difference between the annular pressure reverse order difference and the wellhead pressure reverse order difference of this node is directly used as the pressure propagation deviation range. The annular pressure reverse order difference, the wellhead pressure reverse order difference, and the pressure propagation deviation range together form the reverse order differential response set.

[0065] At the common mass node of 10.0 kg, the annular pressure range of the positive-sequence pressurization trajectory is 11.24 MPa to 11.3 MPa, and the wellhead pressure range is 3.7 MPa to 3.77 MPa; the annular pressure range of the reverse-sequence pressurization trajectory is 11.58 MPa to 11.65 MPa, and the wellhead pressure range is 3.36 MPa to 3.43 MPa. The minimum interval between the annular pressure ranges of the two trajectories is 0.28 MPa, and the minimum interval between the wellhead pressure ranges is 0.27 MPa. Both are greater than the minimum effective non-overlapping amount of pressure direction of 0.25 MPa. The path correlation difference is maintained within a continuous common mass interval of no less than 6.2 kg.

[0066] For each candidate gas injection topology, starting from the corresponding initial state set of the two test cycles, joint set propagation is performed on the forward-order boosting trajectory and the reverse-order boosting trajectory. The two trajectories use the same set of wellbore structure parameters and gas lift valve parameters, and each uses its own process disturbance range. The same parameter is not taken as an independent value in the two trajectories.

[0067] The joint propagation results are converted into an inverse differential reachable domain according to the same effective wellbore quality range, time preimage set, and inverse differential processing as the measured data. When the measured inverse differential response set intersects with the inverse differential reachable domain, the candidate gas injection topology can explain the current quality node; when they do not intersect, the current node is recorded as an inconsistent quality node of the candidate gas injection topology. The changes in external back pressure and wellbore fluid replenishment are already included in the bounded disturbance range of the candidate gas injection topology. Only when the model path without state transition cannot explain the measured inverse differential response within the complete disturbance range, while the model path containing the specified state transition can explain the same response, is the corresponding state transition evidence formed.

[0068] When measured pressure, flow rate, or external operating condition data with valid timestamps exceed the bounded disturbance range, the current topology identification record is marked as insufficient model applicability or measurement inconsistency. The anomaly is not interpreted as downhole gas lift valve switching by excluding a candidate gas injection topology.

[0069] The maximum response delay time used for continuously excluding mass length is taken as the starting point when the effective mass change in the well reaches the effective injection boundary at the wellhead, and as the ending point when the corresponding change in the downhole gas injection channel can be stably reflected in the wellhead pressure measurement. The compressor response, high-pressure gas path propagation and pipe storage changes between the pressure reference command and the effective injection boundary at the wellhead have been corrected by the effective mass coordinates in the well and are not repeatedly accumulated in the maximum response delay time.

[0070] The maximum response delay time is formed by the independent pressure step response records of the gas injection topology that have been confirmed during the commissioning phase. The preferred range for the number of effective records is 6 to 10 sets. In this embodiment, 8 sets are used. For each set of records, the time interval from the moment when the effective injection quality change reaches the effective injection boundary at the wellhead to the time interval when the wellhead pressure enters the stable response range for 3 consecutive unified control cycles is calculated. The maximum value among the upper bounds of the 8 time intervals is taken and superimposed with the upper bound of the time synchronization error to form the maximum response delay time. In this embodiment, the range is 10s to 14s, and 12s is taken. When there are fewer than 6 effective records or the records do not cover the current pressure segment and gas injection topology, a continuous exclusion quality length is not formed, and candidate gas injection topologies are not excluded based on continuous quality intervals. P2 only records the status of unresolved topologies.

[0071] The maximum increment of the effective injection mass upper limit within any 12 seconds is 4.9 kg. After adding twice the mass mapping resolution error, a continuous exclusion mass length of 6.2 kg is obtained. In this embodiment, within the uncertainty range of the flow rate and local mass mapping of the two trajectories, the calculated result of the continuous exclusion mass length is 5.8 kg to 6.8 kg. This embodiment uses the actual calculated value of 6.2 kg. Only the inconsistent mass span is counted for continuous effective common mass nodes. The current inconsistent mass interval is interrupted when a mass node lacking the effective pressure range of any trajectory is missing. When the continuous inconsistent mass span of a candidate gas injection topology reaches 6.2 kg, the candidate gas injection topology is excluded within the current topology identification window.

[0072] The candidate injection topology for "continuous gas flow from the first unloading valve" is excluded because it has a continuous inconsistent mass span of 7 kg within the range of 7 kg to 14 kg. The candidate injection topology for "transition from the first unloading valve to the second unloading valve" cannot explain the pressure recovery after the working valve response is formed within the range of 13 kg to 19.5 kg, and has a continuous inconsistent mass span of 6.5 kg. The candidate injection topology for "continuous gas flow from the working valve" cannot explain the continuous path order difference between the positive and negative pressurization trajectories within the range of 8 kg to 14.4 kg, and has a continuous inconsistent mass span of 6.4 kg. Only the candidate injection topology for "transition from the second unloading valve to the working valve" can simultaneously explain the pressure hysteresis before the transition, the transition interval, and the wellhead pressure recovery after the transition, forming the only valid injection topology for this test pair. For each allowed state transition, the positive and negative pressurization trajectories are checked separately. Only the model path containing the state transition can explain the corresponding response, and the state transition has a wellbore structural basis, forming conditional topology transition evidence B2.

[0073] When calculating the valve depth pressure difference, the pressure on the gas lift valve annulus side minus the pressure on the tubing side is uniformly taken as the positive direction of the valve depth pressure difference. The valve depth pressure on the annulus side is formed by adding the static pressure range of the annulus air column to the surface annulus pressure, and then subtracting the annulus friction range along the gas injection direction. The valve depth pressure on the tubing side is formed by the static pressure range of the mixed fluid in the wellhead filling pipe and the upward flow friction range.

[0074] The annular gas density is obtained using the same version of the pressure, temperature, and gas compressibility factor model as the high-pressure gas path inventory calculation. The range of tubing mixed fluid density, liquid cut range, and friction range are formed by the wellbore structural parameters, the pressure step response of the gas injection topology confirmed during the commissioning phase, and the corresponding surface production records, and are written into the corresponding version of the candidate gas injection topology state model. When there is a lack of valid calibration records for the corresponding pressure segment, no evidence of valve depth differential pressure for that pressure segment is formed.

[0075] Each pressure segment is calculated using the corresponding actual length, temperature range, and flow direction. The valve depth differential pressure evidence interval is taken from the envelope of the candidate model valve depth differential pressure interval that can still explain the corresponding conditional topology transformation evidence. The valve depth differential pressure evidence interval is not pruned by the set of topology transformation parameters to be updated. In this embodiment, the reverse pressurization trajectory forms conditional topology transformation evidence for the closure of the second unloading valve within the effective wellbore mass range of 9.5 kg to 11.5 kg, corresponding to a surface annular pressure of 11.2 MPa to 11.6 MPa and a valve depth differential pressure of 1.28 MPa to 1.42 MPa.

[0076] The working valve opening interval can still be interpreted by multiple state transition branches. Therefore, it is only used as support for the response of the later stage of the candidate gas injection topology transformation of "transformation from the second unloading valve to the working valve" and is not used to update the working valve opening parameters. After the topology identification process is completed, the topology identification record P2 is output. P2 saves the effective gas injection topology set Z2, the conditional topology transformation evidence B2, the reverse order differential response set of each common quality node, the reverse order differential reachability domain identifier of each candidate gas injection topology, the effective cumulative mass range of wells, the common guaranteed coverage quality, and all model and configuration version information. For each gas lift valve opening and closing transformation, a topology transformation parameter set indexed by the gas lift valve identifier, the state before transformation, the state after transformation, and the transformation direction is maintained. Before the new evidence is processed, the submitted topology transformation parameter set is bounded expanded according to the maximum differential pressure drift rate within the same valve type, the same calibration method, and the same pressure and temperature range.

[0077] The maximum differential pressure drift rate is formed by no less than 6 independent and valid calibration records, preferably 6 to 12 records. This embodiment uses 8 records. Each record covers the pressure and temperature range applicable to the current topology conversion parameters. The normal operation update interval is preset by the equipment parameter update configuration version and remains unchanged. If there are fewer than 6 valid records, the record time span cannot cover at least 2 normal operation update intervals, or the valve type and calibration method used by different records are inconsistent, the online shrinkage of the topology conversion parameters will not be performed.

[0078] In this embodiment, the design differential pressure range for the second unloading valve closing transition is 1.1 MPa to 1.6 MPa, the submitted topology transition parameter range is 1.22 MPa to 1.46 MPa, and the preferred range for the maximum differential pressure drift rate is 0.0003 MPa / h to 0.0008 MPa / h. In this embodiment, the maximum differential pressure change per unit operating time from eight adjacent valid calibration records is rounded up to 0.0005 MPa / h. After accumulating 40 hours of operation, the expanded topology transition parameter range is 1.2 MPa to 1.48 MPa. The first cycle, consisting of the 37th and 40th operating cycles... The reverse-order boost test pair forms P2, providing evidence of the second unloading valve closing differential pressure from 1.28 MPa to 1.42 MPa. The second reverse-order boost test pair, consisting of the 44th and 47th running cycles, forms an independent P2, providing evidence of the second unloading valve closing differential pressure from 1.31 MPa to 1.45 MPa. The two P2s do not share test cycles, recovery control segments, or identification data, and use the same model and configuration version. The intersection of the two differential pressure evidence ranges is 1.31 MPa to 1.42 MPa, which, after intersection with the expanded topology transformation parameter range, remains 1.31 MPa to 1.42 MPa.

[0079] The preferred range for resolvable valve depth differential pressure evidence is 0.08 MPa to 0.12 MPa. In this embodiment, 0.1 MPa is used. This value is formed by adding 0.03 MPa for ground pressure measurement error, 0.04 MPa for valve depth differential pressure segmented back-calculation repeatability error, 0.02 MPa for fixed topology model residual, and 0.01 MPa for numerical set outer packaging error. The width of the candidate topology transformation parameter range relative to the previously submitted range is reduced from 0.24 MPa to 0.11 MPa, a reduction of 0.13 MPa. Valve depth differential pressure evidence greater than 0.1 MPa is acceptable. To determine the resolution range, a new second unloading valve closing transition parameter range of 1.31 MPa to 1.42 MPa is submitted. The corresponding test response is marked as a topology transition sample, without entering the channel control parameter update. A separate set of channel control parameters is maintained for each fixed injection channel state. The fixed injection channel states include the first unloading valve overpass state, the second unloading valve overpass state, and the working valve overpass state. The channel control parameters include the preceding annular pressure increment direct gain, the mass domain pressure rise slope action coefficient, and the pressure-time cumulative action coefficient, with dimensions of 1, kg, and s, respectively. -1 .

[0080] The dimension of the annular pressure increment is MPa, which is multiplied by the direct gain to obtain MPa; the dimension of the mass domain pressure rise slope is MPa / kg, which is multiplied by the mass domain pressure rise slope action coefficient to obtain MPa; the dimension of the pressure-time cumulative action is MPa·s, which is multiplied by the pressure-time cumulative action coefficient to obtain MPa. The sum of the three results forms the wellhead pressure increment range. The three parameters have different and fixed physical effects. Only when the effective gas injection topology set Z2 in P2 is a single-element set and the conditional topology transformation evidence B2 is empty, P2 is qualified to update the control parameters within the channel. The positive-sequence pressure increase trajectory and the negative-sequence pressure increase trajectory retain the trajectory source identifier, and the two trajectories are not merged into a data sequence without a source identifier.

[0081] The parameters of the second unloading valve gas passage used for the conversion of the branch are formed by an independent and stable cycle that has been confirmed to be in a continuous gas passage state of the second unloading valve during the commissioning phase. After verification by the same parameter direction separation degree check and non-shared test cycle as the control parameter update in the channel, they are pre-written into the topology control certificate C3. When a valid set of second unloading valve gas passage parameters is not formed, the candidate gas injection topology for "conversion from second unloading valve to working valve" shall not enter the online rolling control. For each actual effective well entry quality node, the preceding annular pressure action time range is determined according to the pressure propagation time range of the candidate gas injection topology, and the annular pressure increment range, the quality domain pressure rise slope range, and the pressure time cumulative action range are formed. The pressure propagation time from the working valve to the wellhead is 8s to 12s. The starting quality node that lacks complete preceding pressure action data shall not enter the control parameter update in the channel.

[0082] Based on the expanded complete set of control parameters within the channel, a scaling scale for parameter directions is formed. For any parameter direction to be distinguished, the change in each parameter in that direction is divided by the projection width of the complete parameter set on the corresponding parameter direction, and the maximum value of the absolute value of the resulting normalized component is made equal to 1. When the projection width of a parameter direction is zero, that parameter is treated as a fixed parameter and does not participate in the construction of the normalized direction. When the projection width of all parameter directions is zero, the current parameter set is maintained and no update of the control parameters within the channel is performed.

[0083] The normalized parameter direction is applied to the annular pressure increment range, the pressure rise slope range, and the cumulative pressure-time range of each quality node, respectively, to form the wellhead pressure change range that the parameter direction may generate. When the change range includes zero, the guaranteed distinguishable response of the quality node to the current parameter direction is set to zero; when the change range does not include zero, the minimum distance from the interval to zero is taken as the guaranteed distinguishable response. The guaranteed distinguishable response is accumulated along the actual effective wellbore mass interval, and then divided by the accumulated value of the response error width of the same quality node. The minimum value of the ratio obtained for all normalized parameter directions is taken to form the parameter direction separation degree.

[0084] The theoretical lower limit of acceptance is 1. Considering the discretization of the quality grid and the ensemble out-of-bounds error, the preferred range of the parameter direction separation threshold is 1.1 to 1.3. In this embodiment, it is 1.2. When the parameter direction separation is not less than 1.2, it is allowed to form a set of parameters to be verified. When the parameter direction separation is less than 1.2, the original set of control parameters in the channel is maintained, and unfavorable quality nodes are not deleted or the response error range is relaxed. The 51st and 54th running cycles form the third reverse-order pressurization test pair. In this test pair, P2 only retains the candidate gas injection topology of "working valve continuous over-gas" and no evidence of conditional topology transformation is formed.

[0085] The preferred number of stable cycles for forming the initial allowable range of the continuous gas flow topology topology for the working valve is 6 to 12. In this embodiment, 8 cycles are used. These 8 stable cycles cover a pressure ramp of 0.04 MPa / s to 0.12 MPa / s and an effective injection mass of 12 kg to 24 kg. For each effective mass node, the three characteristic ranges and the measured wellhead pressure increment range are substituted into the response relationship within the channel to form linear constraints on the three control parameters within the channel. All parameter combinations that allow the calculated wellhead pressure increment range to intersect with the measured wellhead pressure increment range are retained. The feasible set of parameters formed by each mass node is continuously intersected with the initial allowable set of parameters. The initial allowable projection ranges of the three control parameters within the channel are 0.28 to 0.46, 0.6 kg to 1.5 kg, and 0.010 s, respectively. -1 up to 0.032s -1The third reverse-sequence boost test projected the parameters to be verified to range from 0.33 to 0.4, 0.86 kg to 1.18 kg, and 0.015 s, respectively. -1 up to 0.025s -1 The update stops when the feasible set of parameters is empty, and the quality interval with better local fit is not selected for recalculation.

[0086] The set of parameters to be verified is validated by another P2 database that does not share test cycles and recovery control sections, has the same model and configuration version, and is uniquely valid. The maximum drift rate of the three control parameters within the channel is calculated by rounding up the maximum change per unit runtime of adjacent valid identification results in the stable records of the same candidate gas injection topology. Each drift rate is formed using no fewer than 6 independent valid identification records, preferably 8 to 12. If the number of valid records is insufficient, the set of parameters to be verified remains frozen, and online parameter submission is not performed. The preferred range for the direct gain drift rate is 0.0004h. -1 up to 0.0007h -1 In this embodiment, 0.0005h is used. -1 The preferred range for the drift rate of the mass domain pressure ramp effect coefficient is 0.002 kg / h to 0.004 kg / h, and in this embodiment, it is 0.003 kg / h; the preferred range for the drift rate of the pressure-time cumulative effect coefficient is 0.00005 (s·h). -1 Up to 0.0001 (s·h) -1 In this embodiment, the value is taken as 0.00008 (s·h). -1 .

[0087] The 57th and 60th running cycles constitute the fourth reverse-order boost test pair, serving as independent verification records for the set of parameters to be verified formed by the third reverse-order boost test pair. After performing bounded drift expansion based on the cumulative running time of 18 hours between the two records, the projected ranges of the three verification parameters are 0.321 to 0.409, 0.806 kg to 1.234 kg, and 0.0136 s, respectively. -1 up to 0.0264s -1 All of them are narrower than the original submitted parameter projection range. During the verification, the expanded complete set of feasible parameters is used to propagate the wellhead pressure response. The projection ranges of the three parameters are not recombined into independent parameter values. The maximum mass span that is continuously non-intersecting between the predicted wellhead pressure range and the measured wellhead pressure range is statistically predicted along the actual effective wellhead mass coordinate. It is 2kg, which is less than the 6kg continuous exclusion mass length formed by the fourth reverse pressurization test. The verification is successful. The new version of the control parameters in the channel will take effect from the 61st running cycle.

[0088] The third reverse-order boost test pair used to form the set of parameters to be verified is only used as a parameter update record, and the fourth reverse-order boost test pair is only used as a parameter verification record. The fourth reverse-order boost test pair shall not be used as a parameter update record again in the same parameter version. The verified topology transformation parameter set, the complete feasible set of control parameters in the channel corresponding to each fixed gas injection channel state, the projection range of each parameter, the update record identifier, the verification record identifier, the version information, the applicable topology range, and the effective cycle are written into the topology control certificate C3.

[0089] For the candidate gas injection topology of "converting the second unloading valve to the working valve", when the predicted state is in the pre-conversion branch, the set of control parameters in the channel corresponding to the gas passage of the second unloading valve in C3 is called; when entering the post-conversion branch, the set of control parameters in the channel corresponding to the gas passage of the working valve is called. The conversion interval is described by the topology conversion parameter set and the hybrid state branch. The channel control parameters that simultaneously mix the physical effects of the two channels are not established.

[0090] The applicable topology range in C3 only indicates the range of candidate gas injection topologies to which the submitted parameters can be applied. It does not indicate that the next operating cycle has been confirmed to be in a certain gas injection topology. The current candidate gas injection topology in the next operating cycle is still determined by the state observation results. C3 takes effect from the 61st operating cycle. The candidate gas injection topology of "conversion from second unloading valve to working valve" uses the corresponding conversion parameters as well as the gas passage parameters of the second unloading valve before conversion and the gas passage parameters of the working valve after conversion. The candidate gas injection topology of "continuous gas passage of working valve" uses the working valve gas passage parameters newly verified in this embodiment.

[0091] When the 63rd running cycle enters rolling control, the annular pressure is 12.1 MPa and the wellhead pressure is 3.72 MPa. The current candidate gas injection topology set includes two candidate gas injection topologies: "conversion from second unloading valve to working valve" and "continuous gas flow through working valve". Before executing rolling control, it is checked whether all current candidate gas injection topologies are within the parameter applicable topology range of C3, and whether each allowed state branch has corresponding control parameters in the submitted channel. If any current candidate gas injection topology or allowed state branch lacks submitted parameters, the candidate gas injection topology is deleted without set intersection operation, and adaptive boosting optimization is stopped, and the recovery access check is entered.

[0092] In this embodiment, both current candidate gas injection topologies and their allowed state branches have corresponding valid parameters. Therefore, as a parallel prediction scenario, a unified control cycle of 0.5s is adopted. The effective prediction time of the candidate gas injection topology model is determined by independent stable cycles that do not participate in the current parameter update. The preferred range for the number of independent stable cycles is 6 to 10. In this embodiment, 8 cycles are used. The ensemble prediction is started from each control time of each stable cycle, and the longest period during which the measured annular pressure and wellhead pressure are continuously within the prediction range is counted. The minimum value among the longest periods obtained from the 8 stable cycles is taken as the effective prediction time of the model. In this embodiment, the effective prediction time of the model is 30s.

[0093] The preferred range for the prediction time domain is 15s to 25s, with a lower limit greater than the maximum response delay of 12s and a response evaluation interval of no less than 3s. The upper limit is no more than 30s of effective prediction time for the model. In this embodiment, 20s is used, corresponding to 40 prediction steps. Starting from the current state set of each candidate gas injection topology, the allowed state transition branches are propagated using the same version of the hybrid state model as the state observation and topology identification. The predicted effective cumulative mass range of wellbore injection is formed based on the predicted execution command range and the high-pressure gas path inventory relationship. When the predicted effective cumulative mass range of wellbore injection does not have a non-decreasing path or exceeds the applicable range of the gas property model, the current candidate annular pressure reference sequence is not feasible. The annular pressure, wellhead pressure, and effective cumulative mass range records before the current control time are read and combined with the prediction sequence to form the preceding annular pressure range. The preferred range for the historical record window is 12s to 16s. In this embodiment, 12s is used, with a lower limit equal to the maximum response delay. The historical record only takes data from the current control time and before, and does not use measured data after the current control time to form the current execution command.

[0094] For unconfirmed gas lift valve state transitions, the continuous envelope of the valve depth differential pressure within each 0.5s control interval is calculated. The state transition identifier includes the gas lift valve, the state before transition, the state after transition, and the transition direction. For state transitions triggered by an increase in valve depth differential pressure, the upper boundary of the holding zone before the state transition is taken as the lower boundary of the topology transformation parameter set minus the boundary protection amount. The upper boundary of the valve depth differential pressure envelope within the control interval must not exceed the upper boundary of the holding zone. For state transitions triggered by a decrease in valve depth differential pressure, the lower boundary of the holding zone before the state transition is taken as the upper boundary of the topology transformation parameter set plus the boundary protection amount. The lower boundary of the valve depth differential pressure envelope within the control interval must not be lower than the lower boundary of the holding zone. The preferred range of the boundary protection amount is 0.08MPa to 0.25MPa, which only includes the pressure inner loop tracking error, time discretization error, and set outer envelope error that have not yet been included in the valve depth differential pressure envelope. In this embodiment, the upper boundaries of the three errors are 0.06MPa, 0.04MPa, and 0.05MPa, respectively. Therefore, the boundary protection amount is taken as 0.15MPa.

[0095] Only when all current candidate gas injection topologies allow the same gas lift valve to undergo state transition in the same direction, and all candidate gas injection topologies after the transition have the control parameters submitted within the channel, can the predicted valve depth differential pressure envelope be allowed to enter the corresponding topology transition parameter range and continue to propagate the pre-transition branch and the post-transition branch. The equipment hard constraints include: annular pressure not exceeding 16MPa, wellhead pressure between 2.5MPa and 6MPa, compressor load not exceeding 90%, wellhead gas injection valve opening between 10% and 85%, and annular pressure reference slope within the stable tracking range of the inner pressure loop.

[0096] The preferred range for the upper limit of the wellhead gas injection valve opening change in adjacent control cycles is 3 to 7 percentage points. In this embodiment, 5 percentage points are used. This value is determined by the maximum change in the wellhead gas injection valve action calibration that does not produce significant opening overshoot. The recovery access state set is formed offline using the same version of the candidate gas injection topology model and fixed recovery control program as the current C3. The preferred range for the number of independent recovery tests used to form the recovery access state set for each candidate gas injection topology is 6 to 12 sets. In this embodiment, 8 sets are used. Only the initial state that can enter the reference recovery state set within the recovery verification time, does not cross the equipment safety boundary during the entire recovery process, and can complete the recovery in all 8 sets of repeated tests is included in the recovery access state set. States not covered by the recovery test are not added to the recovery access state set by extrapolation.

[0097] The preferred range for the recovery verification duration is 90s to 150s. In this embodiment, 120s is chosen based on the longest recovery time of 96s plus an execution margin of approximately 20%. All predicted states of each candidate gas injection topology at the end of the prediction time domain should be located within the corresponding recovery admission state set. The 20s prediction time domain is divided into 4 control blocks. The preferred range for the duration of each control block is 4s to 6s. In this embodiment, 5s is chosen. The annular pressure reference slope for each control block is selected from 0, 0.04MPa / s, 0.08MPa / s, and 0.12MPa / s, where 0 indicates maintaining the current annular pressure reference value.

[0098] Four control blocks can form a maximum of 256 candidate annular pressure reference sequences. The candidate sequences are generated step by step according to the order of the control blocks. Each time a control block is added, the current control prefix is ​​input into all current candidate gas injection topologies and their allowed state transition branches, and set propagation is performed using the state transition parameter set, control action parameter set, process disturbance range, and complete feasible set of control parameters within the channel. Control prefixes that violate equipment hard constraints, fail to confirm transition holding constraints, or fail to meet terminal recoverability conditions will not be extended to subsequent control blocks.

[0099] The remaining complete sequences that can be executed by all current candidate gas injection topologies constitute a common executable control sequence set. The target wellhead pressure range is given by the predetermined wellhead pressure setpoint corresponding to the current drainage condition in the field production control system. The online predictive control does not modify the target range based on the current prediction result. In this embodiment, the target wellhead pressure range is 3.8MPa to 4.2MPa. When the common executable control sequence set is not empty, the distance of the farthest endpoint of each predicted wellhead pressure range from the target wellhead pressure range is taken as the worst wellhead response deviation, and the worst wellhead response deviation is minimized first.

[0100] In a sequence where the worst wellhead response deviation does not exceed the sum of the optimal value and the equivalent wellhead response tolerance, the minimum remaining interval of the unconfirmed state transition boundary is maximized; when multiple sequences still exist, the average value of the upper limit of the compressor load corresponding to all candidate gas injection topologies is minimized to form the boost load index.

[0101] The preferred range for the wellhead response equivalent tolerance is 0.06 MPa to 0.1 MPa. In this embodiment, it is 0.08 MPa. This value is determined by adding 0.03 MPa wellhead pressure measurement error and 0.05 MPa boost control resolution margin. In the 63rd operating cycle, the first future annular pressure reference value is 12.16 MPa, corresponding to an actual compressor load command of 62% and a wellhead injection valve opening command of 44%. Under the two candidate injection topologies of "second unloading valve to working valve conversion" and "working valve continuous gas flow", the upper limits of the annular pressure in the next 20 seconds are 13.06 MPa and 12.92 MPa, respectively. The wellhead pressure prediction ranges are 3.84 MPa to 4.18 MPa and 3.88 MPa to 4.14 MPa, respectively. The minimum remaining interval for the unconfirmed state transition boundary is 0.18 MPa.

[0102] The controller executes only the first future annular pressure reference value. After a 0.5s unified control cycle, it reads the new pressure, flow rate, and execution feedback, and re-executes the state observation and common annular pressure reference sequence solution. When the common executable control sequence set is empty, it checks whether the current state sets of all candidate injection topologies are located within the corresponding recovery access state sets, and checks whether the hardware interlocks are latched. If all current state sets meet the recovery access conditions and the hardware interlocks are not latched, the fixed recovery control program is executed and parameter updates are frozen. If any candidate injection topology does not meet the recovery access conditions or the hardware interlocks are latched, adaptive control is stopped and switched to the field fixed safety control program or taken over by the equipment hardware interlocks.

[0103] In the 66th operating cycle, the wellhead pressure baseline rises as the output back pressure data with valid timestamps increases. The control sequence satisfying the wellhead pressure target range causes the predicted endpoint of the candidate gas injection topology for the "second unloading valve to working valve conversion" to leave the recovery access state set. The common executable control sequence set is empty. Upon inspection, the current state sets of both current candidate gas injection topologies are within their corresponding recovery access state sets, and the hardware interlock is not latched. Therefore, a fixed recovery control program is executed, causing the annular pressure to decrease in a controlled manner, and simultaneously freezing the topology conversion parameters and channel control parameter updates. Data during the fixed recovery control period must not enter the parameter update and verification chain, nor update the reference recovery state. The state set, high-pressure gas path configuration parameters or C3, and the P1 and P2 formed after recovery are only used to temporarily narrow down the control scenario within the current recovery identification batch, and are not directly submitted for parameter updates. The preferred range of the maximum number of recovery identifications is 1 to 3 times, which is jointly determined by the number of consecutive recovery actions allowed by the equipment, the remaining time of the current operation, and the maximum time required for a single recovery identification. In this embodiment, the equipment allows 3 consecutive recoverys, and the current remaining operation time only allows 2 complete recovery identifications to be completed. Therefore, the maximum number of recovery identifications is 2 times. After 2 recovery identifications, if the current candidate gas injection topology set still has not formed a true subset of the previous candidate gas injection topology set, active identification and adaptive parameter updates are stopped.

[0104] When both P1 and P2 are valid, and sensor calibration, actuator tracking, high-pressure gas path structure, gas properties, and model version are consistent, and the reverse-order boosting trajectory has topology separation capability, if the candidate gas injection topology still cannot be narrowed down, the downhole gas injection channel is output as pending inspection. If there are abnormalities in pressure or flow measurement, abnormalities in actuator tracking, changes in high-pressure gas path configuration, inconsistent gas property versions, insufficient valid samples for maximum response delay, or insufficient applicability of the candidate gas injection topology model, the surface measurement or control model is output as pending verification. Abnormalities are not directly attributed to the downhole gas injection channel. The control execution record P4 saves the C3 identifier, the current candidate gas injection topology set and allowed conversion branches, the status of the common executable control sequence set, the reason for deleting the candidate control sequence, the identifier of the selected control sequence, the executed annular pressure reference value, the actual compressor load command, the actual wellhead gas injection valve opening command, the worst wellhead response deviation, the minimum remaining interval of the unconfirmed state transition boundary, the boosting load index, the parameter update freeze flag, the hardware interlock status, and the abnormal status.

[0105] P4 with parameter update freeze flag, hardware interlock flag, or model inconsistency flag is only used for control process traceability and status verification, and does not enter the topology transformation parameter or channel control parameter update chain. From run cycles 37 to 66, a total of 4 valid reverse-sequence boosting test pairs were formed. The first two reverse-sequence boosting test pairs determined "second unloading valve to working valve conversion" as the only valid gas injection topology, and repeatedly confirmed the second unloading valve closing differential pressure as 1.31 MPa to 1.42 MPa. The third reverse-sequence boosting test pair formed the set of channel control parameters to be verified for the candidate gas injection topology of "working valve continuous gas flow". The fourth reverse-sequence boosting test pair completed independent verification, and the projected ranges of the three submitted channel control parameters were 0.321 to 0.409, 0.806 kg to 1.234 kg, and 0.0136 s, respectively. -1 up to 0.0264s -1 In the 63rd running cycle, when two candidate gas injection topologies coexist, the first feasible annular pressure reference value is executed. In the 66th running cycle, when the common executable control sequence set is empty, all current candidate gas injection topologies meet the recovery admission conditions, and the hardware interlock is not latched, the program switches to the fixed recovery control procedure.

[0106] When applied to other circulating gas lift wells, the candidate gas injection topology and corresponding parameter range are re-determined based on the corresponding wellbore structure, gas lift valve design parameters, instrument calibration results, high-pressure gas path calibration results, fixed recovery test and pressure tracking test. The data transfer, update permissions and control access relationships between P1, P2, C3 and P4 are kept consistent.

[0107] This invention also discloses an adaptive control system for circulating gas lift boosting parameters, comprising: The state observation module establishes a set of candidate gas injection topologies based on the wellbore structure and the gas lift valve conversion relationship. It uses the state model corresponding to each candidate gas injection topology to constrain the operation data of the circulating gas lift well, forming the current state set of each candidate gas injection topology. The reverse pressurization identification module controls the pressurization actuator to execute the forward pressurization trajectory and the reverse pressurization trajectory when all current state sets are within the reference recovery state set. It corrects the cumulative mass of surface metering based on the changes in the high-pressure gas pipeline inventory, aligns the forward pressurization trajectory and the reverse pressurization trajectory with the effective cumulative mass range of the wellbore, and performs continuous mass interval verification between the reverse differential response set and the reverse differential reachable domain of each candidate gas injection topology to form a topology identification record. The parameter update module updates the topology transformation parameter set based on the conditional topology transformation evidence in the topology identification record, updates the control parameter set within the channel based on the fixed topology response that does not form conditional topology transformation evidence, and verifies the updated parameter set using the topology identification record that does not share test cycles to form a topology control certificate. The rolling control module reads the topology control certificate and each current state set, propagates the candidate annular pressure reference sequence of each candidate gas injection topology in parallel, filters the common annular pressure reference sequence according to control constraints, executes the first reference value of the common annular pressure reference sequence, and re-performs state constraints and sequence filtering based on the operating data of the next control cycle.

[0108] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0109] 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; 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, depending on actual needs.

[0110] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for adaptive control of circulating air lift boosting parameters, characterized in that, include: Obtain the operating data of the circulating gas lift well, establish a set of candidate gas injection topologies according to the wellbore structure and gas lift valve conversion relationship, use the state model corresponding to each candidate gas injection topology to constrain the operating data, and form the current state set of each candidate gas injection topology. When each current state set is within the reference recovery state set, the control boosting actuator executes the forward boosting trajectory and the reverse boosting trajectory with opposite boosting order, forming a reverse boosting pairing record; The surface metering cumulative mass is corrected based on the changes in the high-pressure gas pipeline inventory mass. The effective wellbore cumulative mass range is then aligned with the forward-sequence pressurization trajectory and the reverse-sequence pressurization trajectory to generate a reverse-sequence differential response set. The inverse differential response set is subjected to continuous quality interval verification with the inverse differential reachability domain of each candidate gas injection topology to form a topology identification record; The topology transformation parameter set is updated based on conditional topology transformation evidence, the control parameter set within the channel is updated based on the fixed topology response for which no conditional topology transformation evidence has been formed, and the topology identification record without shared test cycles is used for verification to form a topology control certificate. Read the topology control certificate and each current state set, propagate the candidate annular pressure reference sequence of each candidate gas injection topology in parallel, filter the common annular pressure reference sequence according to the control constraints, execute the first reference value of the common annular pressure reference sequence, and re-perform state constraints and sequence filtering based on the operating data of the next control cycle; If no common annular pressure reference sequence is formed, and all current state sets are within the recovery access state set, then fixed recovery control is executed and parameter updates are frozen. If any current state set is outside the corresponding recovery access state set, then the field fixed safety control procedure is executed.

2. The adaptive control method for circulating air lift boosting parameters according to claim 1, characterized in that, The candidate gas injection topology set is established according to the gas lift valve installation order and permissible conversion relationships; The state constraints include: propagating the corresponding state set using the state model corresponding to each candidate gas injection topology, intersecting the propagation result with the pressure measurement constraints to form the current state set of each candidate gas injection topology, and removing candidate gas injection topologies with empty intersections from the candidate gas injection topology set.

3. The adaptive control method for circulating air lift boosting parameters according to claim 2, characterized in that, The reference recovery state set is formed by the state envelope of multiple stable operating cycles under fixed recovery control; Calculate the bidirectional coverage between the current state sets of corresponding candidate gas injection topologies in two running cycles. When the candidate gas injection topology sets of the two running cycles are consistent and the bidirectional coverage satisfies the pairing condition, execute the forward-order boosting trajectory and the reverse-order boosting trajectory respectively. The forward and reverse pressurization trajectories use the same total duration and ground-based metering mass budget, and execute high and low pressurization in reverse order.

4. The adaptive control method for circulating air lift boosting parameters according to claim 3, characterized in that, The formation of the effective cumulative mass range for well penetration includes: The inventory mass range is calculated based on the pressure, temperature, and effective volume of the gas control body between the flowmeter measurement section and the effective injection boundary at the wellhead. The inventory mass change is generated based on the inventory mass range at the start time of the identification window and the current time. The inventory mass change is used to correct the cumulative mass measured on the ground, and a non-reduction path verification is performed on the effective cumulative mass range injected into the well.

5. The adaptive control method for circulating air lift boosting parameters according to claim 4, characterized in that, The formation of topological identification records includes: A common mass node is set in the overlapping interval of the effective cumulative mass range of the positive-sequence pressurization trajectory and the negative-sequence pressurization trajectory, and a negative-sequence differential response set is formed on the common mass node; Using common wellbore structural parameters and gas lift valve parameters, the forward-order boosting trajectory and the reverse-order boosting trajectory are jointly propagated according to the state model corresponding to each candidate gas injection topology, forming the reverse-order differential reachable domain of each candidate gas injection topology; Compare the inverse differential response set with the corresponding inverse differential reachable domain along the continuous mass interval. Remove candidate gas injection topologies from the candidate gas injection topology set whose continuous inconsistent mass span reaches the continuous exclusion mass length. Write the retained candidate gas injection topologies and conditional topology transformation evidence into the topology identification record.

6. The adaptive control method for circulating air lift boosting parameters according to claim 5, characterized in that, The update of the topology transformation parameter set includes: Maintain a set of topology transformation parameters for each air lift valve, state before transformation, state after transformation, and direction of transformation. Based on the conditional topology transformation evidence in two topology identification records that do not share test cycles, form valve depth differential pressure evidence for the same air lift valve state transformation. The valve depth differential pressure evidence is not trimmed using the set of topology transformation parameters to be updated. Find the intersection of the two valve depth differential pressure evidences with the topology transformation parameter set adjusted according to the parameter drift, and write the intersection result into the topology control certificate.

7. The adaptive control method for circulating air lift boosting parameters according to claim 5, characterized in that, The update of the control parameter set within the channel includes: Select topology identification records that retain only one candidate gas injection topology and do not form conditional topology transformation evidence, and perform constraint intersection on the set of control parameters in the channel based on the fixed topology response; The parameter direction separation degree is calculated based on the difference in wellhead pressure response corresponding to different parameter directions. When the parameter direction separation degree meets the admission conditions, a set of control parameters in the channel to be verified is formed. Verification is performed using a topology identification record that does not share the test cycle with the topology identification record that forms the control parameter set within the channel to be verified, and the verified control parameter set within the channel is written into the topology control certificate.

8. The adaptive control method for circulating air lift boosting parameters according to claim 1, characterized in that, The selection of a common annular pressure reference series includes: The applicable topology scope of the verification topology control certificate includes the candidate gas injection topology set. The candidate annular pressure reference sequence is propagated according to each candidate gas injection topology and its allowed switching branches. The candidate annular pressure reference sequence that cannot be executed is screened out based on the equipment operating boundary, gas lift valve switching boundary and terminal recovery condition. Candidate annular pressure reference sequences are selected in order of increasing wellhead pressure prediction deviation and decreasing gas lift valve switching boundary remaining interval to form a common annular pressure reference sequence.

9. The adaptive control method for circulating air lift boosting parameters according to claim 8, characterized in that, When executing the common annular pressure reference sequence, only the first reference value is issued to the booster actuator, and the state constraints and sequence screening are re-performed in the next control cycle; When no common annular pressure reference sequence is formed, the current state set of each candidate gas injection topology is compared with the corresponding recovery access state set. When each current state set is within the corresponding recovery access state set, fixed recovery control is executed, and the updates of the topology transformation parameter set and the channel control parameter set are frozen. When any current state set is outside the corresponding recovery access state set, the on-site fixed safety control procedure is executed.

10. A recirculating air lift booster parameter adaptive control system, employing the recirculating air lift booster parameter adaptive control method as described in any one of claims 1-9, characterized in that, include: The state observation module establishes a set of candidate gas injection topologies based on the wellbore structure and the gas lift valve conversion relationship. It uses the state model corresponding to each candidate gas injection topology to constrain the operation data of the circulating gas lift well, forming the current state set of each candidate gas injection topology. The reverse pressurization identification module controls the pressurization actuator to execute the forward pressurization trajectory and the reverse pressurization trajectory when all current state sets are within the reference recovery state set. It corrects the cumulative mass of surface metering based on the changes in the high-pressure gas pipeline inventory, aligns the forward pressurization trajectory and the reverse pressurization trajectory with the effective cumulative mass range of the wellbore, and performs continuous mass interval verification between the reverse differential response set and the reverse differential reachable domain of each candidate gas injection topology to form a topology identification record. The parameter update module updates the topology transformation parameter set based on the conditional topology transformation evidence in the topology identification record, updates the control parameter set within the channel based on the fixed topology response that does not form conditional topology transformation evidence, and verifies the updated parameter set using the topology identification record that does not share test cycles to form a topology control certificate. The rolling control module reads the topology control certificate and each current state set, propagates the candidate annular pressure reference sequence of each candidate gas injection topology in parallel, filters the common annular pressure reference sequence according to control constraints, executes the first reference value of the common annular pressure reference sequence, and re-performs state constraints and sequence filtering based on the operating data of the next control cycle.

Images