An intelligent low-carbon reservoir parameter inversion and injection-production optimization system

Abstract

The application belongs to the technical field of oil and gas field development engineering, and discloses an intelligent low-carbon oil reservoir parameter inversion and injection-production optimization system, which comprises a data acquisition and access module, an intelligent oil reservoir parameter inversion module, a low-carbon injection-production collaborative optimization module and a scheme execution and instruction issuing module. The intelligent closed-loop system constructed by fusing data driving and physical models effectively solves the core pain points of the single optimization target, the lack of regional collaboration and the dynamic response capability of the prior art. The creativity of the application lies in that: for the first time, the "carbon footprint" is internalized as an optimization target, realizing the systematic balance between economic benefits and low-carbon emission reduction; through integrated scheduling of multiple sources and multiple sinks and collaborative optimization of multiple fluids, the regional resource utilization efficiency and the overall recovery rate are improved; with the help of real-time inversion and rolling optimization mechanism, adaptive dynamic decision-making capability is formed, and the robustness of the system in dealing with uncertainty is significantly enhanced, providing an innovative solution for intelligent and green development of oil fields.

Description

Technical Field

[0001] This invention relates to the field of oil and gas field development engineering technology, and in particular to an intelligent low-carbon reservoir parameter inversion and injection-production optimization system. Background Technology

[0002] In oilfield development, to maintain formation pressure and improve oil recovery, water or gas injection into the ground is widely adopted. Corresponding reservoir parameter inversion and injection-production optimization technologies aim to identify underground reservoir characteristics based on monitoring data through mathematical modeling and calculation methods, thereby optimizing injection and production strategies. Existing technologies mainly include numerical simulation methods based on physical equations and data-driven machine learning methods that have emerged in recent years. The former establishes a detailed geological model for historical fitting and production prediction, while the latter trains a surrogate model to quickly evaluate development plans or directly optimize injection-production parameters.

[0003] While current reservoir parameter inversion and injection-production optimization technologies have made significant progress, the following shortcomings still exist: 1. Existing optimization systems mostly focus on achieving maximum economic benefits or the highest oil recovery rate as the sole objective, failing to systematically incorporate low-carbon indicators such as energy consumption and carbon emissions during the development process into the optimization objective function; 2. Existing injection and production optimization is usually limited to a single reservoir or block, lacking overall optimization capabilities in regional scenarios such as multi-gas source supply, multi-reservoir coordination, and complex pipeline transportation; 3. Most systems adopt an offline static optimization mode, which makes it difficult to make adaptive adjustments and rolling optimizations based on real-time production dynamics and changes in gas supply, and has limited ability to cope with uncertainties.

[0004] With the increasing demands for global energy transition and low-carbon development, oilfield development faces the dual challenges of enhancing recovery rates and controlling carbon emissions. Traditional optimization models centered on economic benefits are no longer sufficient to meet the needs of green and intelligent development of oil and gas fields in the new era. Therefore, there is an urgent need for an intelligent reservoir management system that can simultaneously take into account economic benefits and low-carbon goals, and has the ability to perform regional collaborative optimization and dynamic adjustment. Summary of the Invention

[0005] This invention aims to provide an intelligent, low-carbon reservoir parameter inversion and injection-production optimization system to address the problems mentioned in the background. By constructing an intelligent closed-loop system that integrates data-driven and physical model integration, this invention effectively solves the core pain points of existing technologies, such as single optimization objectives and a lack of regional coordination and dynamic response capabilities. Its innovation lies in: firstly, internalizing "carbon footprint" as an optimization objective, achieving a systematic balance between economic benefits and low-carbon emission reduction; secondly, improving regional resource utilization efficiency and overall recovery rate through multi-source and multi-sink integrated scheduling and multi-fluid collaborative optimization; and thirdly, leveraging real-time inversion and rolling optimization mechanisms to form adaptive dynamic decision-making capabilities, significantly enhancing the system's robustness against uncertainties, and providing an innovative solution for intelligent and green oilfield development.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A smart, low-carbon reservoir parameter inversion and injection-production optimization system includes: The data acquisition and access module is used to acquire real-time monitoring data from the reservoir production site, supply data from external carbon sources, and logistics data from the pipeline network. The reservoir parameter intelligent inversion module is connected to the data acquisition and access module and is used to dynamically update and correct the physical property parameter field of the target reservoir based on the real-time monitoring data through an artificial intelligence model. The low-carbon injection-production collaborative optimization module is connected to the data acquisition and access module and the reservoir parameter intelligent inversion module, respectively. It is used to receive the supply data, logistics data and the updated physical property parameter field, and perform optimization calculations using a comprehensive objective function that includes economic benefits and low-carbon indicators to generate an optimization scheme that includes carbon source allocation, logistics path and wellhead injection-production parameters. The scheme execution and instruction issuance module is connected to the low-carbon injection and production collaborative optimization module. It is used to convert the optimization scheme into executable control instructions and issue them to the corresponding carbon source supply end, logistics pipeline control end and oilfield injection and production equipment end.

[0007] Preferably, the comprehensive objective function in the low-carbon injection-production collaborative optimization module takes minimizing the regional carbon emission intensity per ton of oil or maximizing the regional net carbon sequestration benefit as one of the core optimization objectives.

[0008] Preferably, the supply data acquired by the data acquisition and access module includes real-time or predicted production, purity, price, and carbon intensity factor of multiple carbon dioxide emission sources.

[0009] Preferably, the low-carbon injection-production collaborative optimization module is configured to coordinate and schedule multiple carbon dioxide emission sources in the optimization calculation to complete the carbon dioxide injection allocation to multiple target reservoirs, and simultaneously optimize the water and carbon dioxide alternating injection strategy within each target reservoir.

[0010] Preferably, the system further includes a real-time carbon footprint calculation module, which is connected to the data acquisition and access module and is used to calculate the total carbon emissions and carbon sequestration from the carbon source to the storage site based on the supply data, logistics data and the optimization scheme.

[0011] Preferably, the artificial intelligence model in the reservoir parameter intelligent inversion module is a conditional generative adversarial network or a physical information neural network, and its output is a set of reservoir physical property parameter fields containing uncertainty quantification.

[0012] Preferably, the low-carbon injection-production collaborative optimization module adopts a deep reinforcement learning algorithm, wherein the surrogate model constructed from the parameter field output by the reservoir parameter intelligent inversion module serves as the simulation environment, and the optimization scheme serves as the action strategy of the intelligent agent.

[0013] Preferably, the control commands issued by the scheme execution and command issuance module include: Order instructions for purchasing a specified amount of carbon source; Instructions for adjusting the opening degree of key valves in the pipeline network; Commands for setting the injection medium, injection rate, and injection pressure for a specified injection well.

[0014] Preferably, the system further includes a dynamic feedback and rolling optimization module, which is connected to the data acquisition and access module and the low-carbon injection and extraction collaborative optimization module. The dynamic feedback and rolling optimization module is used to drive the system to re-execute the inversion and optimization process based on the latest monitoring data according to a preset cycle or triggering conditions, so as to update the optimization scheme.

[0015] The beneficial effects of this technical solution compared to existing technologies are as follows: (1) By taking low-carbon indicators such as carbon emission intensity per ton of oil and net carbon sequestration benefits as core optimization targets, a carbon-economic comprehensive evaluation system was constructed, enabling the system to automatically seek the optimal development scheme under the premise of meeting emission reduction requirements, fundamentally changing the traditional optimization model that is guided by a single economic benefit.

[0016] (2) By coordinating and scheduling multiple gas sources, multiple oil reservoirs and complex pipeline networks, cross-regional integrated optimization of carbon flow and production flow has been achieved. This not only improves the utilization efficiency of CO2 resources and pipeline assets, but also significantly improves crude oil recovery rate and carbon sequestration efficiency through refined water-gas alternating injection and other strategies, achieving optimal resource allocation at the global level.

[0017] (3) Based on the real-time parameter inversion of advanced AI models and the rolling optimization of deep reinforcement learning, the system can continuously update its cognition and adjust its strategy according to the dynamic response of the reservoir and changes in external conditions, which greatly enhances the resilience, response speed and overall intelligence level of the production system, and realizes true intelligent reservoir management. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall system architecture provided by the present invention; Figure 2 This is a schematic diagram of the system workflow provided by the present invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments: Example 1 like Figure 1-2 The intelligent low-carbon reservoir parameter inversion and injection-production optimization system shown includes: The data acquisition and access module is used to acquire real-time monitoring data from the reservoir production site, supply data from external carbon sources, and logistics data from the pipeline network. The reservoir parameter intelligent inversion module, connected to the data acquisition and access module, is used to dynamically update and correct the physical property parameter field of the target reservoir based on real-time monitoring data and artificial intelligence models. The low-carbon injection and production collaborative optimization module is connected to the data acquisition and access module and the reservoir parameter intelligent inversion module, respectively. It is used to receive supply data, logistics data and updated physical property parameter fields, and perform optimization calculations using a comprehensive objective function that includes economic benefits and low-carbon indicators to generate an optimized scheme that includes carbon source allocation, logistics path and wellhead injection and production parameters. The scheme execution and instruction issuance module is connected to the low-carbon injection and production collaborative optimization module. It is used to convert the optimization scheme into executable control instructions and issue them to the corresponding carbon source supply end, logistics pipeline control end and oilfield injection and production equipment end.

[0020] This embodiment defines the basic architecture of the system and four core functional modules, which constitute the most basic scope of protection of this invention.

[0021] 1. Data Acquisition and Access Module: This is the system's "sensory nerves," gathering information from three levels through IoT interfaces, data buses, or APIs: (1) Oil reservoir production site: including the oil production, water production, bottom hole flowing pressure and water cut of each production well; the injection volume (water / CO2) and wellhead pressure of the injection well; and possible downhole sensor data (such as distributed fiber optic temperature / vibration measurement DTS / DAS).

[0022] (2) External carbon sources: refers to industrial CO2 emission sources planned for oil displacement and storage, such as power plants, chemical plants, and steel plants.

[0023] (3) Transport pipeline network: refers to the CO2 transport pipeline network connecting carbon source and oil field. Monitoring points include pipeline pressure, flow rate, temperature and valve status.

[0024] 2. Intelligent Reservoir Parameter Inversion Module: This is the system's "cognitive brain." It uses real-time monitoring data obtained from the production site (such as changes in bottom hole pressure and production fluid composition) to drive the AI ​​model to quickly correct its understanding of underground conditions. Traditional "history fitting" takes weeks or even months, while this module can achieve near real-time dynamic updates and output a high-precision three-dimensional distribution field (i.e., digital twin) of physical property parameters such as permeability and porosity, providing an accurate geological basis for optimization.

[0025] 3. Low-Carbon Injection-Production Co-optimization Module: This is the system's "decision-making center." It receives all information from upstream: supply data (what carbon is available), logistics data (how to deliver it), and physical property parameters (where to inject it). Its core innovation lies in adopting a comprehensive objective function. This function is no longer simply about "maximizing net profit," but rather incorporates "carbon emission costs" or "carbon sequestration revenue" as endogenous variables. Based on this, it solves a complex optimization problem and outputs a globally optimal solution. This solution simultaneously specifies: from which carbon source to purchase how much CO2, through which pipeline, allocated to which injection well in which oil field, and the specific injection parameters of that well.

[0026] 4. Scheme Execution and Command Issuance Module: This is the "motor nerve" of the system. It translates "decision language" such as optimization schemes into "control language" that can be recognized in the industrial field, and ensures that the commands are accurately delivered to the corresponding execution agencies (such as the scheduling system of carbon source enterprises, pipeline SCADA system, and wellhead controller of oil fields), thereby forming a closed loop from decision-making to execution.

[0027] Example 2 The comprehensive objective function in the low-carbon injection-production collaborative optimization module takes minimizing the regional carbon emission intensity per ton of oil or maximizing the regional net carbon sequestration benefit as one of its core optimization objectives.

[0028] This embodiment defines the specific form of the optimization objective, which is a direct manifestation of achieving the core of "low carbon".

[0029] 1. Minimize regional carbon emission intensity per ton of oil: i.e., "the net carbon emissions generated per ton of crude oil produced".

[0030] Net carbon emissions = energy consumption emissions from production processes + implicit emissions from purchased CO2 - CO2 stored in geological deposits.

[0031] Optimizing this metric means that the system will automatically favor the use of low-carbon energy sources, utilize "green" CO2 sources (such as CO2 from biomass energy), and maximize storage efficiency.

[0032] 2. Maximize the net benefits of regional carbon sequestration: This objective focuses more on the economics of CCUS projects.

[0033] Net benefit = Revenue from carbon sequestration (such as carbon trading income and tax incentives) + Increased crude oil sales revenue due to CO2 flooding - Costs of CO2 capture, transportation, and injection.

[0034] Optimizing this indicator can enable the system to maximize profits in a carbon market environment.

[0035] Both of these objective functions can be combined with traditional economic objectives (such as NPV) to form a multi-objective optimization problem, and the optimal balance point can be sought through weight allocation or Pareto front analysis.

[0036] Example 3 The data acquisition and access module obtains supply data including real-time or predicted production, purity, price, and carbon intensity factor of multiple carbon dioxide emission sources.

[0037] This embodiment details the specific content of "supply data," which is a prerequisite for achieving multi-source optimized scheduling.

[0038] 1. Production: The real-time availability of carbon sources or the projected production capacity for a future period determines the supply capacity.

[0039] 2. Purity: The concentration of CO2 affects the capture cost and injection effect (high purity is more conducive to miscible oil displacement).

[0040] 3. Price: Procurement cost is a key input for economic calculations.

[0041] 4. Carbon Intensity Factor: This refers to the original emission intensity corresponding to the production of this portion of CO2 (unit: tons of CO2 / ton of product). For example, the carbon intensity factor of CO2 from coal-fired power plants is much higher than that from natural gas processing plants. This factor is used to accurately calculate the carbon footprint and is the basis for achieving realistic emission reduction assessments.

[0042] Example 4 The low-carbon injection-production collaborative optimization module is configured to coordinate and schedule multiple carbon dioxide emission sources in the optimization calculation to complete the carbon dioxide injection allocation for multiple target reservoirs, and simultaneously optimize the water and carbon dioxide alternating injection strategy within each target reservoir.

[0043] This embodiment illustrates the advanced capabilities of the system at two levels: regional collaboration and process optimization.

[0044] 1. Multi-source and multi-sink scheduling: The system acts like a "carbon resource scheduling center." It not only decides which supplier's CO2 to buy, but also which oil field to deliver it to. For example, when power plant A has a sufficient supply of cheap but low-purity CO2, while chemical plant B has a stable supply of high-priced but high-purity CO2, the system may allocate power plant A's CO2 to heavy oil reservoirs that do not require high purity for immiscible flooding, and allocate chemical plant B's CO2 to light oil reservoirs for miscible flooding, in order to achieve the lowest total cost.

[0045] 2. Water-Gas Alternating Injection Strategy Optimization: In order to prevent premature CO2 gas leakage and improve sweep efficiency, an alternating method of injecting a segment of CO2 followed by a segment of water is often adopted. This module can finely optimize the WAG strategy for each target reservoir and even each injection well, including: circulation cycle, CO2 slug size, water slug size, injection rate, etc. This is a manifestation of the depth of "injection and production optimization", which goes beyond simple total allocation.

[0046] Example 5 The system also includes a real-time carbon footprint accounting module, which is connected to the data acquisition and access module. It is used to calculate the carbon emissions and carbon sequestration volume of the entire chain from carbon source to storage site based on supply data, logistics data and optimization schemes.

[0047] This embodiment adds an independent carbon accounting module, providing measurable, reportable, and verifiable support for the system's low-carbon attributes.

[0048] 1. Scope of accounting: Covers the entire life cycle, including: the implicit emissions of the carbon source itself, the energy consumption emissions of the CO2 transportation process, and the electricity consumption emissions of the injection process.

[0049] 2. Calculation output: Real-time calculation of key indicators such as total carbon emissions, net carbon sequestration (total injection - related emissions), and carbon intensity per ton of oil.

[0050] 3. Functions: This data is not only used for internal optimization decisions, but can also automatically generate carbon accounting reports that comply with international or domestic standards, providing a direct and reliable data foundation for carbon asset trading, green finance certification and government regulation.

[0051] Example 6 The artificial intelligence model in the reservoir parameter intelligent inversion module is a conditional generative adversarial network or a physical information neural network, and its output is a set of reservoir physical property parameters containing uncertainty quantification.

[0052] This embodiment defines the types of advanced AI models used for intelligent inversion and their key output characteristics.

[0053] 1. Conditional Generative Adversarial Network: This model can learn to generate multiple possible reservoir parameter fields that conform to geological statistics under given "conditions" (such as well point production data and seismic attributes). Its output is not a definite value, but a probability distribution or a set of multiple possible realizations, which represents the uncertainty of subsurface knowledge.

[0054] 2. Physical Information Neural Network: This model embeds physical equations describing fluid flow (such as Darcy's law) as constraints into the training of the neural network. Even with sparse data, it can ensure that the inversion results conform to physical laws, thereby improving the reliability and generalization ability of the inversion.

[0055] 3. Uncertainty Quantification: The key value of this module is that it outputs a "set of parameter fields" instead of a single model. This enables downstream optimization decisions to be based on risk. For example, it can require the optimization scheme to still perform well under the "worst-case" geological model, thereby enhancing the robustness of the decision.

[0056] Example 7 The low-carbon injection-production collaborative optimization module adopts a deep reinforcement learning algorithm. The surrogate model constructed from the parameter field output by the reservoir parameter intelligent inversion module serves as the simulation environment, and the optimization scheme serves as the action strategy of the intelligent agent.

[0057] This embodiment defines the advanced AI algorithm used for collaborative optimization and its linkage with the inversion module.

[0058] 1. Deep reinforcement learning: In this scenario, the agent is the optimization decision-maker.

[0059] (1) "Environment" is a fast numerical simulator or surrogate model constructed from the reservoir parameter field obtained by inversion; (2) "Action" is an instruction to adjust injection parameters (such as increasing or decreasing the injection volume); (3) The “reward” is the score calculated based on the comprehensive objective function.

[0060] 2. Workflow: The agent learns a policy network by continuously interacting and experimenting with the environment (simulating production over the next few years in simulation). This network maps complex states (current reservoir conditions, carbon source conditions) to optimal actions (optimization schemes). This method is particularly suitable for handling high-dimensional, sequential decision-making problems and can automatically learn complex nonlinear control laws. Its performance often surpasses traditional gradient optimization or heuristic algorithms. Example 8 The control commands issued by the scheme execution and command issuance module include: Order instructions for purchasing a specified amount of carbon source; Instructions for adjusting the opening degree of key valves in the pipeline network; Commands for setting the injection medium, injection rate, and injection pressure for a specified injection well.

[0061] This embodiment illustrates the specific types of control commands, demonstrating the system's ability to directly control the physical world.

[0062] 1. Procurement Instructions: CO2 purchase orders are automatically generated or triggered through enterprise resource planning or supply chain management system interfaces.

[0063] 2. Pipeline regulation command: Through the monitoring and data acquisition system, the opening degree of the main or branch line valves is automatically adjusted to accurately control the CO2 flow direction and flow rate, and realize dynamic routing.

[0064] 3. Wellhead control commands: Through the oilfield production management system or wellhead intelligent controller, the working mode (water injection / CO2 injection), injection displacement and pressure limit of the injection well can be set remotely, which realizes the connection from macro scheduling to micro execution.

[0065] Example 9 The system also includes a dynamic feedback and rolling optimization module, which is connected to the data acquisition and access module and the low-carbon injection and extraction collaborative optimization module. It is used to drive the system to re-execute the inversion and optimization process based on the latest monitoring data according to a preset cycle or triggering conditions, so as to update the optimization scheme.

[0066] This embodiment adds a dynamic feedback and rolling optimization module, upgrading the system from "open-loop optimization" to "closed-loop intelligence".

[0067] 1. Triggering mechanism: It can be a time period (such as every 24 hours) or an event trigger (such as a sudden change in the water cut of a key well or an interruption in the supply of a carbon source).

[0068] 2. Workflow: When the triggering conditions are met, this module automatically starts a new work cycle: collect the latest data → drive the inversion module to update the underground model → drive the optimization module to recalculate based on the new model and the latest external conditions → generate and issue the updated optimization scheme.

[0069] 3. Beneficial effects: This gives the system adaptability and robustness. It can continuously track the dynamic changes of the reservoir and fluctuations in the external environment, and adjust the strategy in real time to ensure that production always runs on the optimal or suboptimal trajectory, truly realizing "intelligent" management.

[0070] The above descriptions are merely embodiments of the present invention, and common knowledge such as specific technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solutions of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. An intelligent low-carbon reservoir parameter inversion and injection-production optimization system, characterized in that, include: The data acquisition and access module is used to acquire real-time monitoring data from the reservoir production site, supply data from external carbon sources, and logistics data from the pipeline network. The reservoir parameter intelligent inversion module is connected to the data acquisition and access module and is used to dynamically update and correct the physical property parameter field of the target reservoir based on the real-time monitoring data through an artificial intelligence model. The low-carbon injection-production collaborative optimization module is connected to the data acquisition and access module and the reservoir parameter intelligent inversion module, respectively. It is used to receive the supply data, logistics data and the updated physical property parameter field, and perform optimization calculations using a comprehensive objective function that includes economic benefits and low-carbon indicators to generate an optimization scheme that includes carbon source allocation, logistics path and wellhead injection-production parameters. The scheme execution and instruction issuance module is connected to the low-carbon injection and production collaborative optimization module. It is used to convert the optimization scheme into executable control instructions and issue them to the corresponding carbon source supply end, logistics pipeline control end and oilfield injection and production equipment end.

2. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 1, characterized in that: The comprehensive objective function in the low-carbon injection-production collaborative optimization module takes minimizing the regional carbon emission intensity per ton of oil or maximizing the regional net carbon sequestration benefit as one of its core optimization objectives.

3. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 1, characterized in that: The data acquisition and access module obtains supply data including real-time or predicted production, purity, price, and carbon intensity factor of multiple carbon dioxide emission sources.

4. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 3, characterized in that: The low-carbon injection-production collaborative optimization module is configured to coordinate and schedule multiple carbon dioxide emission sources in the optimization calculation to complete the carbon dioxide injection allocation for multiple target reservoirs, and simultaneously optimize the water and carbon dioxide alternating injection strategy within each target reservoir.

5. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 4, characterized in that: The system also includes a real-time carbon footprint calculation module, which is connected to the data acquisition and access module and is used to calculate the total carbon emissions and carbon sequestration from the carbon source to the storage site based on the supply data, logistics data and the optimization scheme.

6. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 1, characterized in that: The artificial intelligence model in the reservoir parameter intelligent inversion module is a conditional generative adversarial network or a physical information neural network, and its output is a set of reservoir physical property parameter fields containing uncertainty quantification.

7. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 6, characterized in that: The low-carbon injection-production collaborative optimization module adopts a deep reinforcement learning algorithm, wherein the surrogate model constructed from the parameter field output by the reservoir parameter intelligent inversion module serves as the simulation environment, and the optimization scheme serves as the action strategy of the intelligent agent.

8. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 1, characterized in that, The control commands issued by the scheme execution and command issuance module include: Order instructions for purchasing a specified amount of carbon source; Instructions for adjusting the opening degree of key valves in the pipeline network; Commands for setting the injection medium, injection rate, and injection pressure for a specified injection well.

9. The intelligent low-carbon reservoir parameter inversion and injection-production optimization system according to claim 1, characterized in that: The system also includes a dynamic feedback and rolling optimization module, which is connected to the data acquisition and access module and the low-carbon injection and extraction collaborative optimization module. It is used to drive the system to re-execute the inversion and optimization process based on the latest monitoring data according to a preset cycle or triggering conditions, so as to update the optimization scheme.

Images