Collaborative optimization method for service life and oil and gas yield of offshore oil underwater production system

Abstract

The invention belongs to the technical field of offshore petrochemical engineering, and particularly relates to a collaborative optimization method for service life and oil and gas yield of an offshore oil underwater production system. The invention provides an offshore oil subsea production system life and oil and gas yield collaborative optimization method, and the method achieves the real-time collaborative optimization control based on life perception through the comprehensive evaluation of the equipment health state and the yield dynamic state. Besides, a steady-state operation and control switching mechanism is further configured, safe transition of the system can still be guaranteed when the health state of the offshore oil underwater production system is deteriorated or operation disturbance occurs, and reliable technical support is provided for improving the control precision and operation toughness of the system. The collaborative optimization method for the service life and the oil and gas yield of the offshore oil subsea production system comprises the following steps: evaluating the health state of the offshore oil subsea production system; predicting the oil and gas yield; carrying out collaborative optimization on the service life and the oil and gas yield of the offshore oil underwater production system; and formulating a steady-state real-time control strategy.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of offshore oil and chemical industry, and particularly relates to a method for life and oil and gas production collaborative optimization of an offshore oil underwater production system. BACKGROUND

[0002] In the development process of offshore oil and gas fields, the offshore oil underwater production system as the key equipment for offshore oil and gas development directly affects the production efficiency and operation safety of oil and gas development. With the increase of service time, important components (such as underwater wellheads, Christmas trees, flow manifolds, etc.) in the production system will have different degrees of performance degradation, thereby affecting the productivity and reliability of the whole system. At present, most of the existing optimization methods can only focus on single target optimization, such as maximizing production or minimizing energy consumption, and ignore the dynamic collaborative relationship between equipment life degradation and productivity. In addition, due to the influence of complex and changeable marine environment and uncertain working conditions of the production system, technical personnel need to respond and stabilize the operation process of the offshore oil underwater production system in real time, so as to realize the stable and reliable output of productivity and the sustainable operation of equipment.

[0003] However, further research shows that the existing optimization method lacks integrated real-time collaborative optimization strategy, and it is difficult to realize the collaborative optimal control of productivity and equipment health status under multi-source constraint conditions. Therefore, it is urgent for the technical personnel in the field to provide a new collaborative optimization method which integrates equipment life evaluation, production dynamic regulation and steady-state real-time optimization, so as to improve the intelligent level and operation efficiency of the offshore oil underwater production system. SUMMARY

[0004] The application provides a method for life and oil and gas production collaborative optimization of an offshore oil underwater production system, which realizes real-time collaborative optimization control based on life perception by comprehensively evaluating the equipment health status and production dynamics of the offshore oil underwater production system. In addition, the collaborative optimization method is also configured with a steady-state operation and control switching mechanism, which can still guarantee the safe transition of the system when the health status of the offshore oil underwater production system deteriorates or the system is disturbed, thereby providing strong technical support for improving the control accuracy and operation resilience of the offshore oil underwater production system.

[0005] To solve the above technical problems, the application adopts the following technical solutions: The method for life and oil and gas production collaborative optimization of an offshore oil underwater production system comprises the following steps: Step S1: constructing a degradation model of the offshore oil underwater production system; dynamically estimating the degradation state of each equipment in the offshore oil underwater production system; and predicting the life of the offshore oil underwater production system; Step S2: determining a key variable affecting oil and gas production; constructing an oil and gas production prediction model to obtain an oil and gas production prediction result; and predicting a sequence of the oil and gas production; Step S3: constructing a life and oil and gas production collaborative optimization objective function of the offshore oil underwater production system, and solving and outputting a multi-objective optimization result; Step S4: performing instruction processing on the control strategy; determining a state feedback mechanism based on a model and data fusion; performing real-time rolling optimization and strategy updating; and establishing a steady-state operation constraint control and switching mechanism of the offshore oil underwater production system.

[0006] Preferably, the process of constructing the degradation model of the offshore oil underwater production system in step S1 is specifically described as follows: A Gamma process model is used to model the degradation characteristics of the offshore oil underwater production system, satisfying: ; wherein S(t) is the degradation state of the equipment at time t; a is a shape parameter, representing a degradation rate; b is a scale parameter, representing a degradation fluctuation; and t is time. The statistical characteristics of the Gamma distribution satisfy: ; wherein D is a failure threshold of the equipment; E[] is a mean value of the distribution; and Var[] is a variance of the distribution. When S(t) is greater than or equal to D, it is determined that the equipment fails.

[0007] Preferably, the process of dynamically estimating the degradation state of each equipment in the offshore oil underwater production system in step S1 is specifically described as follows: A particle filtering algorithm is used to dynamically estimate the degradation state of each equipment in the offshore oil underwater production system. The system state equation satisfies: ; and the observation equation satisfies: ; wherein ; , is observation noise; and Z t is an observation value of a sensor. The particle set is initialized using a prior distribution Each particle is sampled according to state transition to obtain: ; wherein ; The observation likelihood of each particle is calculated according to the observation value, and the weight is updated to obtain: ; wherein R t is an observation noise covariance matrix. The normalized weight is resampled to obtain the state estimation result at the current time, satisfying: .

[0008] More preferably, the process of predicting the service life of the offshore oil underwater production system in step S1 is specifically described as: Since the Gamma process has the independent increment property, the service life of the offshore oil underwater production system satisfies: ; The corresponding cumulative distribution function satisfies: ; Where P() is a probability distribution operation, and FGamma is the cumulative distribution function of the Gamma distribution.

[0009] More preferably, the process of constructing an oil and gas production prediction model to obtain an oil and gas production prediction result in step S2 is specifically described as: Using a random forest regression algorithm to predict the oil and gas production in the future period; Assuming that the input feature vector at the current time t is: ; The constructed oil and gas production prediction model satisfies: ; Where f RF is a random forest regression function, and ε t is a residual term; Collecting the historical data set {(x i , Q i )} i=1 N Including the input variables and actual production at each time, and performing normalization processing; Randomly generating M regression trees, each tree performing sampling with replacement from the training samples, each tree node only selecting the optimal partition feature from the feature subset, and independently training each tree to output a sub-prediction value ; Taking the output of each regression tree as the final oil and gas production prediction result, obtaining: .

[0010] More preferably, the process of constructing a service life and oil and gas production collaborative optimization objective function of the offshore oil underwater production system in step S3, solving and outputting the multi-objective optimization result is specifically described as: Taking maximizing the oil and gas production and minimizing the equipment life loss in the future rolling time window as the collaborative optimization objective, the constructed collaborative optimization objective function satisfies: ; Where J1 and J2 are objective functions, u t and u kis the control variable, H is the control sequence, C d () represents the life consumption cost function; The life consumption in unit time satisfies: ; The double objectives in the collaborative optimization objective function are combined into a weighted objective function, and the following is obtained: ; Wherein, g1+g2=1, g1 and g2 are target weights; if maximizing oil and gas production is preferred, then g1>g2; if minimizing equipment life consumption is preferred, then g1<g2; The simulated annealing algorithm is used to globally search and optimize the constructed weighted objective function; the initial control sequence is U(0), the initial temperature is Te0, the termination temperature is Te min , and the cooling factor b is set to be in (0, 1); In each iteration, a new candidate control sequence U new is generated based on the current control sequence, satisfying: ; Wherein, U current is the control sequence at the current time, and ΔU is a preset small change value; The objective function value J new of the candidate solution is calculated, satisfying: J new =J(U new ); If J new <J current , then the solution is accepted; otherwise, the inferior solution is accepted with the following probability: ; Wherein, Te is the current temperature; The search optimization process is terminated until the temperature is lower than the termination temperature or the iteration number reaches the upper limit.

[0011] The application provides a life and oil and gas production collaborative optimization method for a marine oil underwater production system, comprising the following steps: constructing a degradation model of the marine oil underwater production system; dynamically estimating the degradation state of each device in the marine oil underwater production system; predicting the life of the marine oil underwater production system; determining key variables affecting oil and gas production; constructing an oil and gas production prediction model to obtain an oil and gas production prediction result; predicting the sequence of oil and gas production; constructing a life and oil and gas production collaborative optimization objective function for the marine oil underwater production system, solving and outputting the multi-objective optimization result; performing instruction processing on the control strategy; determining a state feedback mechanism based on the model and data fusion; performing real-time rolling optimization and strategy updating; and establishing a steady-state operation constraint control and switching mechanism for the marine oil underwater production system.

[0012] The marine oil underwater production system life and oil and gas production collaborative optimization method with the step characteristics has at least the following technical advantages compared with the prior art: (1) The marine oil underwater production system life and oil and gas production collaborative optimization method provided by the application realizes real-time collaborative optimization control based on life perception driving by fusing the health state and production dynamics of underwater equipment; the degradation level of key equipment in the marine oil underwater production system is effectively quantified based on life modeling and health index evaluation mechanism, and the future production trend is evaluated in combination with a prediction model, so that more constraint and forward-looking decision basis is provided for the control strategy; and the heuristic algorithms such as rolling optimization and simulated annealing are adopted to efficiently search the control strategy under the conditions of multiple constraints and multiple variables, so that the dynamic balance of the production increase and life extension of the marine oil underwater production system is realized. (2) The marine oil underwater production system life and oil and gas production collaborative optimization method provided by the application can still guarantee the safe transition and recovery of the system under the health state deterioration or operation disturbance of the marine oil underwater production system by setting the steady-state operation domain and the control switching mechanism, so that the control precision, economic benefit and operation resilience of the marine oil underwater production system are improved, and the method has significant practical value. BRIEF DESCRIPTION OF DRAWINGS

[0013] The accompanying drawings are used to provide a further understanding of the application, and constitute a part of the specification, together with the embodiments of the application, to explain the application, and do not constitute a limitation of the application. In the following drawings: Figure 1 The flowchart of the marine oil underwater production system life and oil and gas production collaborative optimization method provided by the application; Figure 2 The degradation state curve diagram obtained; Figure 3 The life loss prediction curve diagram; Figure 4 The oil and gas production prediction result and actual execution production comparison diagram; Figure 5 The simulated annealing optimization target value descent curve diagram; Figure 6 The control variable adjustment strategy diagram; Figure 7 The comprehensive benefit change diagram before and after the marine oil underwater production system life and oil and gas production collaborative optimization method provided by the application is executed. DETAILED DESCRIPTION

[0014] The application provides a kind of marine oil underwater production system life and oil and gas production collaborative optimization method, which realizes real-time collaborative optimization control based on life perception by comprehensively evaluating equipment health status and production dynamics in marine oil underwater production system;In addition, the collaborative optimization method is also configured with steady-state operation and control switching mechanism, which can still guarantee the safe transition of the system when the health status of marine oil underwater production system deteriorates or operating disturbance occurs, providing strong technical support for improving the control accuracy and operating resilience of marine oil underwater production system.

[0015] The application provides a kind of marine oil underwater production system life and oil and gas production collaborative optimization method, as shown in Figure 1 The steps include: Step S1: Constructing the degradation model of marine oil underwater production system; dynamically estimating the degradation state of each equipment in marine oil underwater production system; predicting the life of marine oil underwater production system.

[0016] It is worth noting that before implementing other steps, the application first evaluates the health status of marine oil underwater production system. Specifically, as a more preferred embodiment of the application, the process of constructing the degradation model of marine oil underwater production system in this step S1 is specifically described as follows: The Gamma process model is used to model the degradation characteristics of marine oil underwater production system, which satisfies: ; Where S(t) is the degradation state of the equipment at time t; α is the shape parameter, representing the degradation rate; β is the scale parameter, representing the degradation fluctuation; t is the time; The statistical properties of Gamma distribution satisfy: ; Where E[] is the mean of the distribution, Var[] is the variance of the distribution; D is the failure threshold of the equipment; when S(t) ≥ D, it is determined that the equipment fails.

[0017] During the process, it needs to be noted that by constructing the degradation model of offshore oil underwater production system and describing its performance degradation process over time, it provides a theoretical basis for subsequent steps of state estimation and life prediction. Specifically, by collecting the operation and maintenance log data during the operation of the equipment (which includes various data such as cumulative running time of the equipment, maintenance records at each time point, replacement time of key components, failure time and corresponding failure mode), according to the failure time and cumulative running time, a time-failure contrast relationship is established (which can be used to describe the information when the degradation state reaches the failure threshold). Then, the key parameter data collected by each underwater sensor is collected, such as temperature, pressure, vibration intensity, flow, torque, current and voltage signals. After processing by feature extraction algorithm, a monotonically increasing degradation feature sequence reflecting the degree of equipment performance degradation can be obtained. In addition, by collecting the performance evaluation data of the equipment in the actual operation process (which includes quantitative indicators reflecting the performance decline trend of the equipment such as liquid production rate change rate, start-stop response time delay, unit cycle energy consumption, system overall efficiency), and normalizing and time aligning the above quantitative indicators, the feature variables that increase with time can be extracted, which can be used as the observation input of the degradation state.

[0018] On the basis of completing the construction of the degradation model of offshore oil underwater production system, the degradation state of each equipment in the offshore oil underwater production system is further dynamically estimated. It is worth noting that based on real-time observation data, the degradation state of each equipment in the offshore oil underwater production system is dynamically estimated by fully considering factors such as data noise, system nonlinearity and working condition uncertainty, which can realize online tracking of the health state of the equipment.

[0019] Specifically, particle filtering algorithm is used to dynamically estimate the degradation state of each equipment in the offshore oil underwater production system. Among them, the Gamma process model constructed in the foregoing steps is used as the state transition model framework, and three types of heterogeneous observation data (i.e. operation and maintenance log data, sensor monitoring data and performance index data) are used for state correction.

[0020] The system state equation satisfies: The observation equation satisfies: ; Wherein, ; , belongs to observation noise; Z t is the observation value of the sensor.

[0021] It is additionally pointed out that the Z tThe observation value of the sensor contains the fusion result of the degradation characteristic value obtained from the sensor, the performance index sequence value and the state related label data extracted from the operation and maintenance log. Further research can find that the observation value Z of the sensor t Specifically composed of the following sub-data vectors: that is, Z t (1) The real-time degradation features such as vibration acceleration and flow rate drop amplitude extracted from the sensor monitoring data; Z t (2) The dynamic performance change indicators such as response time and fluid production rate in the performance index data; Z t (3) The state events or maintenance labels such as state warnings and minor failure records in the operation and maintenance log data corresponding to the current time window.

[0022] By defining the following fusion function, the observation mapping relationship can be constructed: ; Where H1, H2, H3 are the mapping matrices corresponding to the three types of observation data, ω1, ω2, ω3 ∈ [0, 1] are weight coefficients, ω1+ ω2+ ω3=1, which can be dynamically adjusted according to experience.

[0023] Then, the particle set is initialized by the prior distribution For each particle, state transition sampling is performed to obtain: ; Where, .

[0024] According to the observation value, the observation likelihood of each particle is calculated, and the weight is updated to obtain: .

[0025] Where, R t is the observation noise covariance matrix; The weights are normalized and resampled to obtain the state estimation result at the current time, which satisfies: .

[0026] On the basis of completing the dynamic estimation of the degradation state of each device in the offshore oil underwater production system, the life of the offshore oil underwater production system is further predicted. It is worth noting that since the Gamma process has the independent increment property, the life of the offshore oil underwater production system satisfies: .

[0027] The corresponding cumulative distribution function satisfies: ; Where P() is the probability distribution operation, and FGamma is the cumulative distribution function of the Gamma distribution.

[0028] Step S2: Determine the key variables that affect oil and gas production; construct an oil and gas production prediction model to obtain oil and gas production prediction results; and predict the oil and gas production sequence.

[0029] Here, step S2 is specifically used to predict oil and gas production. It is worth noting that by identifying and determining the key variables that affect oil and gas production, a (data) input basis can be provided for building subsequent prediction models.

[0030] Further research revealed that the key variables affecting oil and gas production can be divided into the following three categories: (1) Equipment health status variables: such as the life span RUL of the current equipment t and the estimated degradation state of the current equipment; (2) Operating parameter variables: including but not limited to inlet pressure p in,t , outlet pressure p out,t , wellbore temperature T t , fluid density ρ t and moisture content W t； (3) Control parameter variables: for example, the speed v of the downhole electric submersible pump t and nozzle opening k t , the electric submersible pump speed v t and nozzle opening k t The parameters can be uniformly expressed as u t =[v t k t ] T And, the output variable can be expressed as the oil and gas production Q of the current period t .

[0031] In addition, as a preferred embodiment of the present invention, the process of constructing the oil and gas production prediction model and obtaining the oil and gas production prediction result in step S2 is specifically described as follows: Use the random forest regression algorithm to predict oil and gas production in the next cycle; Assume that the input feature vector at the current time t is: ; The constructed oil and gas production prediction model satisfies: ; Among them, f RF is the random forest regression function, ε t is the residual term; Collect historical data sets {(x i ,Q i )} i=1 N The input variables and actual output at each moment included in are normalized; Randomly generate M regression trees, each tree is sampled from the training samples with replacement, each tree node only selects the optimal partition feature from the feature subset, and each tree is independently trained, and the sub-prediction value is output ; The outputs of each regression tree are averaged as the final oil and gas production prediction result, and the following is obtained .

[0032] Step S3: Constructing a life and oil and gas production collaborative optimization objective function of the offshore oil underwater production system, solving and outputting the multi-objective optimization result.

[0033] Among them, the step S3 is specifically used for collaborative optimization of the life and oil and gas production of the offshore oil underwater production system. It should be noted that the main targets of collaborative optimization are the following two key indicators: maximizing oil and gas production and minimizing equipment life loss in the future rolling time window. Therefore, the collaborative optimization objective function constructed by taking maximizing oil and gas production and minimizing equipment life loss in the future rolling time window as the collaborative optimization target meets: ; Wherein, J1 and J2 are objective functions, u t and u k are control quantities, H is a control sequence, C d () represents a life loss cost function; The life loss per unit time meets: .

[0034] Then, after obtaining the collaborative optimization objective function, a set of optimal control strategies meeting the engineering constraint conditions is solved by a multi-objective optimization algorithm.

[0035] Specifically, the double targets in the collaborative optimization objective function are combined into a weighted objective function, which is: ; Wherein, g1+g2=1, g1 and g2 are target weights; if maximizing oil and gas production is preferred, then g1>g2; if minimizing equipment life loss is preferred, then g1<g2; The simulated annealing algorithm is used to globally search and optimize the constructed weighted objective function; the initial control sequence is set as U(0), the initial temperature is set as Te0, the termination temperature is set as Te min , and the cooling factor b∈(0,1) is set; In each iteration, a new candidate control sequence U new is generated based on the current control sequence, which meets: ; Wherein, U current is the control sequence at the current time, and ΔU is a preset small change value; Computing the objective function value J of the candidate solution new , satisfying: J new = J(U new ); If J new < J current , then accept the solution; otherwise accept the inferior solution with a probability: ; where Te is the current temperature. The search optimization process is terminated until the temperature is lower than the termination temperature or the number of iterations reaches an upper limit.

[0036] It should be noted that, in view of the actual operation characteristics of the offshore oil underwater production system, the system operating environment has the characteristics of deep water high pressure, high corrosion, high cost, long period, and the like, and there is a complex coupling relationship between the core devices, and the control strategy needs to consider both device degradation suppression and production efficiency improvement. In the actual optimization process, the collected data at least includes: the residual life prediction value output by the device health assessment, the real-time production curve, the wellhead and manifold pressure, the pump speed, the bottom hole flowing pressure, the energy consumption index, the device start-stop frequency, the control valve state sequence, and the like. Therefore, the collaborative optimization objective function constructed based on the above multi-source data should at least cover two core objectives: maximizing the unit time production and minimizing the cumulative life loss of key devices (such as electric submersible pumps, SCM control valve groups, etc.).

[0037] In addition, it should be additionally pointed out that the optimal control strategy output by the above simulated annealing algorithm can be expressed as which can be specifically used to guide the generation of real-time control strategies in subsequent steps. The purpose of doing so not only meets the multi-objective coordinated optimization requirements of oil and gas production and device life, but also can be directly used to guide the subsequent real-time operation control, ensuring that the underwater production system realizes the comprehensive operation goal of reasonable energy consumption, stable device, and high production efficiency in the extreme marine environment.

[0038] Step S4: instructing the control strategy; determining the state feedback mechanism based on the model and data fusion; real-time rolling optimization and strategy updating; establishing the steady-state operation constraint control and switching mechanism of the offshore oil underwater production system.

[0039] The step S4 is specifically used to formulate a steady-state real-time control strategy.

[0040] Specifically, instructing the control strategy means mapping the control vector output by the optimization into executable control instructions of the execution unit of the offshore oil underwater production system. The control vector includes standardized pump speed instructions, valve opening degree instructions, and other key control quantities.

[0041] Further, the optimized standardized pump speed value is mapped to actual voltage and current control signals through a motor drive model to control the operating frequency and start-up curve of the downhole electric submersible pump. The valve opening percentage is converted into an angular displacement signal of the actuator through a nonlinear position-angle conversion model to control key fluid control elements such as the choke valve, throttle valve, and gas lift valve.

[0042] The determination of the state feedback mechanism based on the model and data fusion refers to real-time acquisition of the current control input, current observation output, current degradation state prediction value, and current life prediction value in each control cycle; then, based on the model and data fusion (feedback condition), the yield prediction trend (error) and life loss condition are re-evaluated to finally generate the state feedback mechanism.

[0043] The real-time rolling optimization and strategy update refers to further adopting a rolling horizon control method, in which the control rolling window is moved forward in each control cycle, and then the health assessment and yield prediction model input are updated.

[0044] Among them, only the control quantity at the first optimization time is applied to the system, and the rest is reserved as an alternative, and the above process is repeated to form a closed-loop control system, ensuring that the marine oil underwater production system of the present application can still operate stably and sustainably under the conditions of health state degradation, external disturbance, or sensor drift.

[0045] The establishment of the steady-state operation constraint control and switching mechanism of the marine oil underwater production system refers to further establishing a steady-state operation constraint control and switching mechanism for the marine oil underwater production system. Among them, due to the operation characteristics of long distance, unattended, high pressure and high corrosion, strong environmental disturbance, etc., the core equipment of the system includes wellhead Christmas tree, underwater control module, underwater pump system, multiphase flow metering device, and valve actuator, etc. During operation, by collecting real-time control data including pump speed, current, voltage, valve opening, wellhead pressure, liquid production, etc., as well as health index, residual life estimate value output by the equipment health assessment model, and key physical quantities such as temperature, vibration, and flow monitored by the sensor, continuous monitoring and analysis of the system operation state are realized.

[0046] During the optimization and control strategy execution process, the system continuously assesses whether it is operating within the preset steady-state operating range based on the aforementioned multi-source data. This means that production is within the target range, system efficiency is above the threshold, equipment degradation is under control, and there are no potential risk warning indicator violations. If monitoring results indicate that the system is operating outside this steady-state range, such as abnormal production fluctuations, pump current overload, or the health index falling outside the safe range, the control switching mechanism is immediately triggered, and the control strategy is corrected online, automatically adjusting the control parameters of key equipment. For example, the speed of the submersible pump and the range of the compression nozzle opening are adjusted downward, and the valve opening and closing frequency is limited to quickly return the system to the steady-state operating range. This avoids equipment damage, severe production fluctuations, or operational failures caused by control imbalance, thereby ensuring the safety and continuity of deep-sea oil and gas production.

[0047] To facilitate those skilled in the art to understand the present invention, a set of simulation production optimization examples are further provided herein to verify the effectiveness and reliability of the method for collaborative optimization of the life span and oil and gas production of marine oil underwater production systems provided by the present invention.

[0048] Specifically, first provide various parameters required in the simulation process. The names and values ​​of the parameters can be referred to in the following table.

[0049] ; Then, a degradation model of the offshore oil underwater production system is constructed. Figure 2 As shown in the figure, the degradation state curve is generated by using the Gamma process model; the quantitative indicators are normalized and time-series aligned, where the normalized life loss index Cd is obtained to represent the decline of equipment health over time, as shown in the figure. Figure 3 shown.

[0050] Then, an oil and gas production prediction model is constructed to predict oil and gas production. Specifically, a production prediction model Q is constructed that couples control variables (pump speed, valve opening) with degradation status and environmental factors. t , capturing the combined effect of the system operating status on oil and gas production. The comparison chart of the oil and gas production prediction results and the actual production can be referred to as follows: Figure 4 shown.

[0051] Then, considering the dual objectives of production and life, the multi-objective optimization results are solved and output. Figure 5 As shown, Figure 5 The simulated annealing optimization target value decrease curve is shown. By optimizing the control variable sequence through simulated annealing, the objective function (negative output plus weighted loss) is minimized, thus obtaining the optimal control strategy for pump speed and valve.

[0052] Then, a state feedback mechanism based on model and data fusion is determined. When the equipment health index is lower than the safety threshold, the corresponding emergency adjustment mechanism is triggered to gently reduce the control variable to reduce further life consumption. The specific control variable adjustment strategy is shown in FIG. Figure 6

[0053] Finally, the benefit change before and after the execution of the life and oil and gas production collaborative optimization method of the offshore oil underwater production system provided by the present application is compared and analyzed. Referring to FIG. Figure 7

[0054] In summary, the above examples show through numerical simulation that the life and oil and gas production collaborative optimization method of the offshore oil underwater production system provided by the present application, how the optimization strategy realizes the adaptive optimization of the control variable (pump speed, valve opening), and how the feedback mechanism realizes the emergency adjustment, thereby improving the comprehensive benefit of the offshore oil underwater production system.

[0055] The present application provides a life and oil and gas production collaborative optimization method of an offshore oil underwater production system, comprising the following steps: constructing a degradation model of the offshore oil underwater production system; dynamically estimating the degradation state of each device in the offshore oil underwater production system; predicting the life of the offshore oil underwater production system; determining the key variable affecting oil and gas production; constructing an oil and gas production prediction model to obtain an oil and gas production prediction result; predicting the sequence of oil and gas production; constructing a life and oil and gas production collaborative optimization objective function of the offshore oil underwater production system, solving and outputting the multi-objective optimization result; instructing the control strategy; determining a state feedback mechanism based on model and data fusion; real-time rolling optimization and strategy updating; establishing a steady-state operation constraint control and switching mechanism of the offshore oil underwater production system.

[0056] The life and oil and gas production collaborative optimization method of the offshore oil underwater production system has at least the following technical advantages compared with the prior art: ​​(1) The marine oil underwater production system life and oil and gas production collaborative optimization method provided by the present application realizes real-time collaborative optimization control based on life perception driving by fusing the health state and production dynamic of underwater equipment; based on life modeling and health index evaluation mechanism, the degradation level of key equipment in the marine oil underwater production system is effectively quantified, and the future production trend is evaluated in combination with a prediction model, thereby providing more constraint and forward-looking decision basis for the control strategy; rolling optimization and heuristic algorithms such as simulated annealing are adopted to efficiently search for the control strategy under the conditions of multiple constraints and multiple variables, and the dynamic balance between the production improvement and life extension of the marine oil underwater production system is realized; (2) The marine oil underwater production system life and oil and gas production collaborative optimization method provided by the present application can still guarantee the safe transition and recovery of the system to stability under the condition of deterioration of the health state of the marine oil underwater production system or operation disturbance by setting a steady-state operation domain and a control switching mechanism, and therefore has significant practical value in improving the control precision, economic benefit and operation resilience of the marine oil underwater production system.

[0057] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for collaboratively optimizing the lifespan and oil and gas production of an offshore oil underwater production system, characterized in that: The following steps are included: Step S1: constructing a degradation model of an offshore oil underwater production system; dynamically estimating the degradation state of each device in the offshore oil underwater production system; and predicting the life of the offshore oil underwater production system; Step S2: Determine the key variables that affect oil and gas production; construct an oil and gas production prediction model to obtain oil and gas production prediction results; and predict the oil and gas production sequence; Step S3: constructing a coordinated optimization objective function for the lifespan and oil and gas production of an offshore oil underwater production system, and solving and outputting the multi-objective optimization results; Step S4: Instruction-based control strategy processing; determine the state feedback mechanism based on model and data fusion; Real-time rolling optimization and strategy update; Establish steady-state operation constraint control and switching mechanism for offshore oil underwater production systems; The process of constructing the degradation model of the offshore oil underwater production system in step S1 is specifically described as follows: The Gamma process model is used to model the degradation characteristics of offshore oil underwater production systems to meet the following requirements: ; Where S(t) is the degradation state of the device at time t; α is the shape parameter, which represents the degradation rate; β is the scale parameter, which represents the degradation fluctuation; t is time; The statistical characteristics of the Gamma distribution satisfy: ; Where E[] is the mean of the distribution, Var[] is the variance of the distribution; D is the failure threshold of the equipment; When S(t)≥D, the device is considered to be failed; The process of constructing the oil and gas production prediction model and obtaining the oil and gas production prediction result in step S2 is specifically described as follows: Use the random forest regression algorithm to predict oil and gas production in the next cycle; Assume that the input feature vector at the current time t is: ; The constructed oil and gas production prediction model satisfies: ; Among them, f RF is the random forest regression function, ε t is the residual term; Collecting historical datasets The input variables and actual output at each moment included in are normalized; Randomly generate M regression trees, each tree is sampled with replacement from the training sample, each tree node selects only the optimal partitioning feature from the feature subset, and trains each tree independently to output the sub-prediction value ; The output of each regression tree is averaged as the final oil and gas production prediction result, and we get: .

2. The method for collaboratively optimizing the lifespan and oil and gas production of an offshore oil underwater production system according to claim 1, characterized in that: The process of dynamically estimating the degradation status of each device in the offshore oil underwater production system in step S1 is specifically described as follows: Using particle filtering algorithm, the degradation status of each device in the offshore oil underwater production system is dynamically estimated; The system state equation satisfies: ; The observation equation satisfies: ; in, ; , which belongs to observation noise; Z t is the observation value of the sensor; Initialize the particle set using prior distribution , sample each particle according to state transition and get: ; in, ; According to the observed value, the observation likelihood of each particle is calculated and the weight is updated to obtain: ; Among them, R t is the observation noise covariance matrix; Normalize the weights and resample to obtain the state estimation result at the current moment, which satisfies: .

3. The method for collaboratively optimizing the lifespan and oil and gas production of an offshore oil underwater production system according to claim 1, characterized in that: The process of predicting the life of the offshore oil underwater production system in step S1 is specifically described as follows: Since the Gamma process has the property of independent increments, the life of the offshore oil underwater production system satisfies: ; The corresponding cumulative distribution function satisfies: ; Where P() is the probability distribution operation, and FGamma is the cumulative distribution function of the Gamma distribution.

4. The method for collaboratively optimizing the lifespan and oil and gas production of an offshore oil underwater production system according to claim 1, characterized in that: In step S3, the objective function of collaborative optimization between the life span and oil and gas production of the offshore oil underwater production system is constructed, and the multi-objective optimization results are solved and outputted. The specific description is as follows: Taking maximizing oil and gas production and minimizing equipment life loss in the future rolling time window as the collaborative optimization goals, the collaborative optimization objective function is constructed to meet the following requirements: ; Among them, J1 and J2 are objective functions, u t and u k is the control quantity, H is the control sequence, C d () represents the life loss cost function; The life loss per unit time satisfies: ; The dual objectives in the collaborative optimization objective function are combined into a weighted objective function to obtain: ; Among them, g1+g2=1, g1 and g2 are target weights; if maximizing oil and gas production is the priority, then let g1>g2; if minimizing equipment life loss is the priority, then let g1 <g2; The simulated annealing algorithm is used to perform global search optimization on the constructed weighted objective function; the initial control sequence is U(0), the initial temperature is Te0, and the termination temperature is Te min , set the cooling factor b∈(0,1); In each iteration, the current control sequence is perturbed to generate a new candidate control sequence U new ,satisfy: ; Among them, U current is the control sequence at the current moment, ∆U is the preset small change value; Calculate the objective function value J of the candidate solution new , meet: J new =J(U new ); If J new <J current , then accept the solution; otherwise accept the inferior solution with the following probability: ; Where Te is the current temperature; The search optimization process is terminated when the temperature is lower than the termination temperature or the number of iterations reaches the upper limit.

Images