Hydropower flow region dispatching method, device and equipment and storage medium

Abstract

The application discloses a water and electricity basin scheduling method and device, equipment and storage medium, relates to the technical field of information processing, and comprises the following steps: acquiring water and electricity basin scheduling business scope and hierarchical management information, constructing a multi-level agent cooperation architecture based on an access control protocol, determining the decision boundary and interaction rules of each agent, wherein, according to the preset water and electricity field professional corpus, a low-rank adaptation and retrieval enhancement double mechanism is adopted to provide scheduling reasoning support for each agent; acquiring basin hydrological monitoring data, power generation constraint data and flood control safety data, driving the multi-level agent to perform collaborative reasoning, obtaining a plurality of target reasoning results, and through the main agent, the target reasoning results are fused and decided to obtain a decision fusion result and reasoning basis information; a penetrating cognitive decision display interface is constructed to realize cross-level penetrating display and instruction issuing of decision information. The application can realize accurate and efficient, clear rights and responsibilities, and can be explained and supervised integrated intelligent scheduling.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to a method, apparatus, device and storage medium for hydroelectric domain scheduling. Background Technology

[0002] Currently, the main technical solution in the field of watershed scheduling is a combination of distributed information systems and single intelligent decision-making units. General large models are beginning to be applied to the scheduling inference process. Some solutions have achieved cross-level data transmission and basic visualization, forming the current technology application pattern with data aggregation and simple scheduling calculation as the core.

[0003] However, existing technologies have obvious shortcomings. Hydrological, power generation, and flood control data are scattered and independent, making them difficult to access in a unified manner. A single intelligent agent cannot cover the needs of multi-objective decision-making. The lack of professional knowledge in the hydropower industry in general large models leads to poor adaptability of dispatch instructions inference. Cross-level decision information is only transmitted as data without penetrating the reasoning process. The lack of clear decision boundaries in multi-agent collaboration can easily lead to task conflicts and imbalances in multi-objective optimization. Ultimately, this results in delayed response and insufficient accuracy in flood control and power generation collaborative decision-making. At the group level, there is a lack of effective supervision of plant dispatch execution. Summary of the Invention

[0004] The main purpose of this application is to provide a watershed scheduling method, device, equipment and storage medium, which aims to effectively solve the problems of scattered watershed scheduling data, poor adaptability of general models, chaotic multi-agent collaboration, insufficient cross-level decision penetration and imbalance of multiple objectives of flood control and power generation, so as to achieve integrated intelligent scheduling that is accurate, efficient, clear in rights and responsibilities, and interpretable and superviseable.

[0005] To achieve the above objectives, this application proposes a hydroelectric domain scheduling method, the method comprising: The system acquires the scope and hierarchical management information of hydropower domain scheduling business, constructs a multi-level intelligent agent collaborative architecture based on access control protocol, and determines the decision boundaries and interaction rules of each intelligent agent. In particular, based on the preset hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is adopted to provide scheduling reasoning support for each intelligent agent. The system acquires watershed hydrological monitoring data, power generation constraint data, and flood control safety data, drives multi-level intelligent agents to perform collaborative reasoning, obtains several target reasoning results, and then uses the main intelligent agent to perform fusion decision-making on each of the target reasoning results to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result. A penetrating cognitive decision-making display interface is constructed, which pushes the decision fusion result and the corresponding reasoning basis information to the management personnel, so as to realize the cross-level penetrating display of decision information and the issuance of instructions.

[0006] In one possible implementation, the steps of acquiring the scope and hierarchical management information of hydropower domain scheduling services, constructing a multi-level intelligent agent collaborative architecture based on an access control protocol, and determining the decision boundaries and interaction rules of each intelligent agent include: Based on the scope of the hydroelectric current domain scheduling business and the hierarchical management information, the hierarchical division information is determined; Based on the hierarchical division information and the access control protocol, the decision boundaries and data interaction permissions of each intelligent agent are determined; Based on the decision boundaries and data interaction permissions, establish interaction rules and collaboration mechanisms for multi-level intelligent agents.

[0007] In one possible implementation, establishing multi-level agent interaction rules and collaboration mechanisms based on the decision boundary and the data interaction permissions includes: Obtain the task type and execution scope corresponding to each of the aforementioned intelligent agents; Based on the task type and the execution scope, determine the task flow and result aggregation method among the intelligent agents; Based on the decision boundaries, data interaction permissions, and task flow and result aggregation methods, a multi-level intelligent agent collaborative workflow is formed.

[0008] In one possible implementation, the provision of scheduling and reasoning support for each intelligent agent, based on a pre-set hydropower domain corpus and employing a dual mechanism of low-rank adaptation and retrieval enhancement, includes: Acquire the watershed operation knowledge and scheduling procedures corresponding to the professional corpus in the hydropower field; For any of the aforementioned intelligent agents, based on the knowledge of the watershed operation and the scheduling procedure, a low-rank adaptation process is performed on the basic large model. A knowledge base is constructed based on the results of the low-rank adaptation process, and retrieval enhancement is performed to form a scheduling reasoning support capability.

[0009] In one possible implementation, the acquisition of watershed hydrological monitoring data, power generation constraint data, and flood control safety data drives multi-level intelligent agents to perform collaborative reasoning, obtaining several target reasoning results. A main intelligent agent then performs a fusion decision on each of the target reasoning results to obtain the fusion decision result and the reasoning basis information corresponding to the fusion decision result, including: Based on the hydrological monitoring data, the hydrological intelligent agent is driven to generate hydrological prediction and inference results. Based on the power generation constraint data, the power generation agent is driven to generate power generation optimization inference results. Based on the flood control safety data, the flood control intelligent agent is driven to generate flood control safety inference results; Based on the hydrological prediction reasoning results, the power generation optimization reasoning results, and the flood control safety reasoning results, the main intelligent agent performs multi-objective fusion to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result.

[0010] In one possible implementation, the step of obtaining the decision fusion result and the corresponding reasoning basis information by the main intelligent agent through multi-objective fusion based on the hydrological prediction reasoning result, the power generation optimization reasoning result, and the flood control safety reasoning result includes: Obtain information on the current reservoir operating water level and capacity constraints in the watershed; Based on the reservoir operating water level and the reservoir capacity constraint information, the basin inflow trend and water allocation constraints are determined according to the hydrological prediction reasoning results; the unit power generation output and power generation duration allocation scheme is determined according to the power generation optimization reasoning results; and the reservoir safe water level and discharge flow constraints are determined according to the flood control safety reasoning results. Multi-objective balance fusion is performed to generate decision fusion results under conventional scheduling scenarios and the reasoning basis information corresponding to the decision fusion results.

[0011] In one possible implementation, the construction of a penetrating cognitive decision-making display interface, which pushes the decision fusion result and the corresponding reasoning basis information to management personnel, enables cross-level penetrating display of decision information and issuance of instructions, includes: Obtain the display elements and risk warnings corresponding to the decision fusion results and the reasoning basis information; Based on the displayed elements and the risk warnings, a penetrating cognitive decision-making display interface is constructed. The decision fusion results are simulated and the effects are previewed using a watershed digital twin model, and then synchronized to the penetrating cognitive decision-making display interface. Based on the penetrating cognitive decision display interface, decision information is pushed to management personnel, and cross-level instructions and manual intervention are completed.

[0012] Furthermore, to achieve the above objectives, this application also proposes a hydroelectric current domain scheduling device, which includes: The acquisition module is used to acquire the business scope and hierarchical management information of hydropower domain scheduling, construct a multi-level intelligent agent collaborative architecture based on access control protocol, and determine the decision boundary and interaction rules of each intelligent agent. Among them, based on the preset hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is adopted to provide scheduling reasoning support for each intelligent agent. The decision module is used to acquire watershed hydrological monitoring data, power generation constraint data and flood control safety data, drive multi-level intelligent agents to perform collaborative reasoning, obtain several target reasoning results, and perform fusion decision on each of the target reasoning results through the main intelligent agent to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result. The push module is used to build a penetrating cognitive decision display interface, push the decision fusion result and the reasoning basis information corresponding to the decision fusion result to the management personnel, and realize the cross-level penetrating display of decision information and the issuance of instructions.

[0013] In addition, to achieve the above objectives, this application also proposes a hydroelectric current domain scheduling device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the hydroelectric current domain scheduling method as described above.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the hydroelectric domain scheduling method described above.

[0015] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the hydroelectric current domain scheduling method described above.

[0016] This application provides a method, apparatus, device, and storage medium for hydropower domain scheduling. The method acquires the scope and hierarchical management information of hydropower domain scheduling services, constructs a multi-level intelligent agent collaborative architecture based on an access control protocol, and determines the decision boundaries and interaction rules of each intelligent agent. Specifically, based on a pre-set hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is employed to provide scheduling reasoning support for each intelligent agent. This allows for the acquisition of basin hydrological monitoring data, power generation constraint data, and flood control safety data, driving the multi-level intelligent agents to perform collaborative reasoning, obtaining several target reasoning results, and transmitting them through… The main intelligent agent performs fusion decision-making on the reasoning results of each target, and obtains the fusion decision result and the reasoning basis information corresponding to the fusion decision result. This constructs a penetrating cognitive decision display interface, and pushes the fusion decision result and the reasoning basis information corresponding to the fusion decision result to the management personnel. This enables cross-level penetrating display of decision information and instruction issuance, thereby effectively solving the problems of scattered watershed scheduling data, poor adaptability of general models, chaotic multi-agent collaboration, insufficient cross-level decision penetration, and imbalance between multiple objectives of flood control and power generation. This achieves integrated intelligent scheduling that is accurate, efficient, clear in responsibilities, interpretable, and superviseable. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating an embodiment of the hydroelectric current domain scheduling method of this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the hydroelectric current domain scheduling method of this application; Figure 3 This is a schematic diagram of the module structure of the hydroelectric current domain scheduling device according to an embodiment of this application; Figure 4 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the hydroelectric current domain scheduling method in this application embodiment.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device, big data service platform, or hydroelectric domain scheduling system capable of realizing the above functions. The following description uses a hydroelectric domain scheduling system as an example to illustrate this embodiment and the subsequent embodiments.

[0024] Based on this, this application provides a method for water current domain scheduling, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the hydroelectric current domain scheduling method of this application.

[0025] In this embodiment, the hydroelectric current domain scheduling method includes steps S11 to S13: Step S11: Obtain the scope and hierarchical management information of hydropower domain scheduling business, construct a multi-level intelligent agent collaborative architecture based on access control protocol, and determine the decision boundary and interaction rules of each intelligent agent. In this step, according to the preset hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is adopted to provide scheduling reasoning support for each intelligent agent. It should be noted that the scope of hydropower domain scheduling business refers to the business area and content covered by hydrological monitoring, power generation scheduling, and flood control within the hydropower domain; hierarchical management information refers to the management structure and management responsibilities of hydropower domain scheduling from the group level to the plant level; access control protocol refers to the communication protocol used to regulate the data interaction and decision execution permissions of intelligent agents; multi-level intelligent agent collaboration architecture refers to the collaborative working architecture composed of multiple intelligent agents with different functions; intelligent agent refers to a functional unit with independent data processing and decision reasoning capabilities; decision boundary refers to the business scope in which each intelligent agent can independently execute decisions; interaction rules refer to the constraint rules for data transmission and task collaboration between intelligent agents; hydropower professional corpus refers to professional data containing hydropower domain operation knowledge, cascade reservoir scheduling procedures, and power generation equipment operation standards; low-rank adaptation refers to a lightweight training method that injects industry knowledge into the basic large model; retrieval enhancement refers to a processing method based on knowledge base for semantic matching and knowledge retrieval; and scheduling reasoning support refers to providing intelligent agents with decision reasoning capabilities that conform to hydropower industry rules.

[0026] In this embodiment, by acquiring the scope and hierarchical management information of watershed scheduling services, the system can clarify the management architecture and business coverage of watershed scheduling. Based on the access control protocol, a multi-level intelligent agent collaborative architecture is constructed, which can clearly define the decision boundaries and data interaction permissions of each intelligent agent, avoiding task conflicts and permission overstepping between intelligent agents. At the same time, combined with the preset professional corpus of hydropower, the system adopts a dual mechanism of low-rank adaptation and retrieval enhancement to strengthen the reasoning ability of intelligent agents, enabling the decision reasoning of intelligent agents to conform to the actual scheduling needs of the hydropower industry and improve the accuracy and adaptability of the reasoning results.

[0027] Specifically, in one possible implementation, the system uses an access control list protocol as the underlying access control protocol to achieve fine-grained control over the permissions of intelligent agents. Furthermore, when constructing a multi-level intelligent agent collaborative architecture, the system can classify intelligent agent types according to the functional dimensions of hydrological prediction, power generation optimization, and flood control decision-making, and simultaneously classify intelligent agent levels according to the management dimensions of the group, river basin, and power plant, thus forming a complete three-level collaborative architecture. For example, the system acquires the business scope of the Wujiang cascade reservoir scheduling and the hierarchical management information from the group to the power plant, constructs a three-level intelligent agent collaborative architecture for hydrology, power generation, and flood control based on the access control list protocol, clarifies the decision-making scope and data interaction permissions of each intelligent agent, and collects professional corpora such as the hydrological characteristics of the Wujiang River basin and the scheduling procedures of the cascade reservoirs. It employs a dual mechanism of low-rank adaptation and retrieval enhancement to provide each intelligent agent with reasoning support that conforms to the scheduling rules of the Wujiang River basin, ensuring that the intelligent agents can carry out reasoning work based on the actual situation of the basin.

[0028] Step S12: Obtain watershed hydrological monitoring data, power generation constraint data and flood control safety data, drive multi-level intelligent agents to perform collaborative reasoning, obtain several target reasoning results, and perform fusion decision on each of the target reasoning results through the main intelligent agent to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result; It should be noted that: watershed hydrological monitoring data refers to monitoring data reflecting watershed precipitation, flow, water level, and other hydrological conditions; power generation constraint data refers to constraint data used to limit the power output and duration of generating units; flood control safety data refers to relevant data such as water level, reservoir capacity, and outflow used to ensure the flood control safety of reservoirs; multi-level intelligent agents refer to a collection of intelligent agents with different scheduling functions and belonging to different management levels; collaborative reasoning refers to the process by which multiple intelligent agents cooperate to complete data processing and decision calculation; target reasoning results refer to the reasoning conclusions generated by each intelligent agent for its own responsible target; the master intelligent agent refers to the core intelligent agent used to coordinate the results of various intelligent agents and execute fusion decision; fusion decision refers to the process of integrating the reasoning results of multiple sub-targets into a unified scheduling scheme; decision fusion result refers to the final scheduling scheme that takes into account multiple objectives of flood control and power generation; and reasoning basis information refers to the calculation basis and rule basis that support the generation of decision fusion results.

[0029] In this embodiment, by acquiring watershed hydrological monitoring data, power generation constraint data, and flood control safety data, the system can provide a complete data foundation for agent reasoning, drive multi-level agents to perform collaborative reasoning, and enable different agents to complete the calculation and analysis of corresponding sub-objectives. By integrating the reasoning results of each objective through the main agent, the system can resolve the multi-objective optimization conflict between flood control and power generation, obtain a decision fusion result that takes into account both safety and efficiency, and generate corresponding reasoning basis information to support subsequent decision display and supervision.

[0030] In one possible implementation, the system employs a reactive reasoning framework to drive multi-level agents to complete the collaborative reasoning process. Furthermore, when executing fusion decisions, the main agent bases its decisions on reservoir operating water level and capacity constraints, balancing various conditions such as inflow trends, power generation allocation, and flood control limitations to ensure that the decision results comply with the basin's routine scheduling requirements.

[0031] In one specific implementation, the system acquires real-time hydrological monitoring data, generator power generation constraint data, and reservoir flood control safety data of the Wujiang River basin. It then drives the hydrological agent, power generation agent, and flood control agent to complete inflow prediction, power generation optimization, and flood control safety calculation, respectively, and obtain the corresponding target reasoning results. The main agent combines the current reservoir operating water level and reservoir capacity constraints to perform multi-objective balance fusion of the three types of reasoning results, and obtains the decision fusion result of the routine scheduling of the Wujiang cascade reservoirs. Simultaneously, it generates the reasoning basis information corresponding to the result.

[0032] Step S13: Construct a penetrating cognitive decision display interface, push the decision fusion result and the reasoning basis information corresponding to the decision fusion result to the management personnel, and realize the cross-level penetrating display of decision information and the issuance of instructions.

[0033] It should be noted that the "penetrating cognitive decision-making display interface" refers to a visual operation interface that enables cross-level data drill-down and decision-making process display. "Management personnel" refers to staff responsible for watershed scheduling approval and supervision. "Cross-level penetrating display" refers to the full-link display of decision information from the group level to the plant level. "Instruction issuance" refers to the operation process of pushing scheduling decisions to the field execution terminal. In this embodiment, the system constructs a penetrating cognitive decision-making display interface that can present the decision fusion results and reasoning basis information in a visual form, allowing management personnel to intuitively view scheduling plans and decision reasons. This achieves cross-level penetrating display of decision information and pushes decision instructions to the field, enabling rapid implementation of scheduling plans. It also supports manual intervention and adjustment of decisions by management personnel, solving the problem in traditional scheduling where decision information is only transmitted without explanation and the group lacks a basis for supervision, thus improving the interpretability and supervisory capacity of scheduling decisions.

[0034] In one possible implementation, the system combines a watershed digital twin model to simulate and preview the results of decision fusion, and synchronizes the simulation results to the display interface. Additionally, the system can push decision information and dispatch instructions to on-site dispatchers via mobile terminals, enabling real-time, penetrating dissemination of decision instructions.

[0035] In one specific implementation, the system constructs a penetrating cognitive decision-making display interface for the Wujiang River Basin, which fully displays the decision-making fusion results and reasoning basis information of cascade reservoir scheduling on the interface. At the same time, the scheduling plan is simulated and pre-run through the Wujiang River Basin digital twin model and synchronized to the interface. Group and plant management personnel can view the four-level data drilling information and risk warnings through the interface. The system pushes the final scheduling instructions to the on-site scheduling personnel through mobile terminals, completing the cross-level decision-making penetration display and instruction issuance.

[0036] For the complete implementation process, taking the daily scheduling and extreme rainfall early warning scenario of the Wujiang cascade reservoirs as an example, the system executes the following process in sequence: 1. The system acquires the business scope of the Wujiang cascade reservoir scheduling and the three-level hierarchical management information of the group, basin, and power plant. Based on the access control protocol, it constructs a collaborative architecture of hydrological intelligent agent, power generation intelligent agent, flood control intelligent agent, and master intelligent agent, clarifying the decision-making boundaries and data interaction permissions of each intelligent agent. At the same time, it acquires professional corpus in the hydropower field, such as the hydrological characteristics of the Wujiang basin, the scheduling procedures of the cascade reservoirs, and the operating standards of the power plants. It uses a dual mechanism of low-rank adaptation and retrieval enhancement to train the basic large model in an industry-specific manner, and builds a basin-specific knowledge base to provide scheduling reasoning support for each intelligent agent.

[0037] 2. The system acquires real-time hydrological monitoring data of the Wujiang River Basin, including the current average daily inflow into the reservoir, reservoir operating water level, and rainfall data in the area; acquires power generation constraint data, including maximum and minimum output of generating units and constraints on the number of operating units; and acquires flood control safety data, including flood control limit water level, maximum allowable discharge flow, and safe storage capacity constraints in the reservoir area.

[0038] 3. The system drives the hydrological intelligent agent to carry out predictive reasoning based on hydrological monitoring data, and generates the inflow trend and water volume prediction results for the next 72 hours; it drives the power generation intelligent agent to generate the unit output optimization and power generation time allocation results based on power generation constraint data; and it drives the flood control intelligent agent to generate the reservoir water level control and discharge flow control results based on flood control safety data.

[0039] 4. The system acquires real-time status information such as the current operating water level and regulation capacity constraints of the Wujiang cascade reservoirs. Combining hydrological forecast results, power generation optimization results, and flood control safety results, the main intelligent agent performs multi-objective balance fusion: Under normal scheduling scenarios, the system controls the reservoir water level within a reasonable range below the flood limit, allocates unit output according to the optimal range, and ensures power generation benefits while meeting safety constraints; when an extreme heavy rainfall warning is issued in the basin, the system automatically switches to a flood control priority strategy, lowers the reservoir operating water level in advance, reserves flood control capacity, appropriately reduces power generation output, and strictly controls the downstream flow to ensure the safety of the reservoir area and downstream river channels. Finally, it generates scheduling decision schemes and corresponding reasoning information under extreme weather conditions.

[0040] 5. The system extracts display elements such as decision-making schemes, water levels, flow rates, power output, constraints, and risk warnings to construct a penetrating cognitive decision-making display interface. It also uses a digital twin model of the Wujiang River Basin to simulate and extrapolate the scheduling scheme, predicting the water level changes, flow processes, and power output trends over the next 72 hours, and synchronizes the simulation results to the display interface.

[0041] 6. The system pushes decision-making schemes, reasoning basis, risk warnings, and simulation results to managers at all levels through a display interface. Managers can view, confirm, and manually adjust these information. Finally, dispatch instructions are pushed across levels to the power plant's on-site execution terminal, enabling the dispatch instructions to be issued penetratingly and to be monitored and traceable throughout the entire process.

[0042] This embodiment effectively solves the technical problems of scattered watershed scheduling data, poor adaptability of general models, chaotic multi-agent collaboration, insufficient cross-level decision penetration, and imbalance of multi-objective optimization by constructing a multi-level intelligent agent collaborative architecture with clear permissions, enhancing industry-adaptive reasoning capabilities, completing multi-objective fusion decision-making for flood control and power generation, and building an interpretable and superviseable penetrating display interface. It achieves efficient, accurate, and supervised integrated intelligent scheduling of the watershed.

[0043] In one feasible implementation, the steps of acquiring the scope and hierarchical management information of hydropower domain scheduling services, constructing a multi-level intelligent agent collaborative architecture based on an access control protocol, and determining the decision boundaries and interaction rules of each intelligent agent include: Step S21: Determine the hierarchical division information based on the scope of the hydroelectric current domain scheduling business and the hierarchical management information; It should be noted that the hierarchical classification information refers to the multi-level scheduling hierarchy classification results formed based on business and management dimensions. In this embodiment, the system classifies the hierarchy based on the scope of hydropower domain scheduling business and hierarchical management information, which can fit the actual scheduling management system and provide a unified hierarchical benchmark for subsequent intelligent agent deployment and permission allocation, ensuring that the multi-level intelligent agent architecture is consistent with the actual business organizational architecture. In one possible implementation, the system determines the hierarchical classification information according to a three-level management model of group level, watershed level, and plant level.

[0044] Specifically, the system first analyzes the scope of hydropower basin scheduling operations, extracting key information such as the number of reservoirs covered, basin areas, and scheduling types. Then, it combines this information with hierarchical management information, including management responsibilities, control levels, and organizational structures, classifying and integrating the data according to a combination of management levels and business types. This results in hierarchical information suitable for multi-level intelligent agent deployments. For example, based on the coverage of the Wujiang cascade reservoir scheduling operations and the hierarchical management information of China Huadian Corporation, the system determines a three-level hierarchical structure including the group's management level, the basin centralized control level, and the power plant execution level.

[0045] Step S22: Determine the decision boundaries and data interaction permissions of each intelligent agent based on the hierarchical division information and the access control protocol; It should be noted that access control protocols refer to communication control protocols used to constrain the data access and decision-making permissions of intelligent agents. The decision boundary of an intelligent agent refers to the business scope and hierarchical range within which each agent can independently conduct reasoning and decision-making. The data interaction permissions of an intelligent agent refer to the data types and data ranges that each agent can read, send, and access. In this embodiment, the system combines hierarchical division information and access control protocols to assign exclusive decision boundaries and data interaction permissions to each intelligent agent. This prevents agents from making decisions or accessing data beyond their authority, reduces the probability of task conflicts between agents, and improves collaborative stability. In one possible implementation, the system uses an access control list protocol to achieve precise allocation and constraint of agent permissions.

[0046] Specifically, the system determines the management level and business direction of each agent based on the hierarchical classification information. Then, according to the access control protocol, it sets the decision-making matters that the agent can execute, the data resources that can be invoked, and the collaborative requests that can be initiated, ultimately clarifying the decision-making boundaries and data interaction permissions corresponding to each agent. For example, based on the three-level hierarchical classification information of the Wujiang River Basin and the access control list protocol, the system assigns corresponding business decision-making boundaries and corresponding types of data interaction permissions to the hydrological agent, the power generation agent, and the flood control agent.

[0047] Step S23: Based on the decision boundary and the data interaction permissions, establish the interaction rules and collaboration mechanism of multi-level intelligent agents.

[0048] It should be noted that interaction rules refer to the behavioral norms for data transmission, task notification, and result feedback between intelligent agents, while the collaboration mechanism refers to the overall working mode of multiple intelligent agents completing task decomposition, parallel reasoning, and result aggregation. In this embodiment, the system establishes standardized interaction rules and collaboration mechanisms based on decision boundaries and data interaction permissions, enabling intelligent agents to cooperate in an orderly and efficient manner, achieving unified task distribution, independent process execution, and unified result aggregation, forming a complete scheduling decision-making closed loop. In one possible implementation, the system establishes request-response interaction rules between intelligent agents according to a task-driven model.

[0049] Specifically, the system determines the task processing scope based on the decision boundaries of each agent, determines the data transmission path and method based on data interaction permissions, and on this basis, formulates communication methods, task triggering conditions, result reporting methods, and exception handling methods between agents, ultimately forming a stable and reliable multi-level agent interaction rule and collaboration mechanism. For example, based on the decision boundaries and data interaction permissions of each agent, the system establishes data transmission rules, task invocation rules, and result aggregation processes among the three levels of agents in the Wujiang River Basin, forming a stable and efficient multi-level agent collaborative working mechanism.

[0050] This embodiment clarifies the division of labor, scope of authority, and collaboration methods of multi-level intelligent agents by sequentially completing the hierarchical division, permission allocation, interaction rules, and collaboration mechanism establishment. It solves the technical problems of chaotic multi-agent collaboration, unclear decision boundaries, and easy task conflicts, and improves the overall collaborative efficiency and decision reliability of watershed scheduling.

[0051] In one feasible implementation, establishing multi-level agent interaction rules and collaboration mechanisms based on the decision boundary and the data interaction permissions includes: Step S31: Obtain the task type and execution scope corresponding to each of the intelligent agents; It should be noted that an intelligent agent refers to a functional unit with independent data processing and decision-making reasoning capabilities; task type refers to the category of scheduling business handled by each intelligent agent; and execution scope refers to the business area and management level coverage that each intelligent agent can carry out business processing. In this embodiment, the system obtains the task type and execution scope corresponding to each intelligent agent, which can clearly define the business responsibilities and processing boundaries of each intelligent agent, providing a basis for subsequent task flow and collaborative work, and ensuring that each intelligent agent performs its own duties without business overlap or omission. In one possible implementation, the system divides task types into hydrological prediction tasks, power generation optimization tasks, flood control decision-making tasks, and main intelligent agent coordination tasks.

[0052] Specifically, based on the functional design of the multi-level intelligent agent collaborative architecture, the system extracts the business categories undertaken by each intelligent agent, and determines the area and hierarchical scope that the intelligent agent can perform scheduling processing by combining the hierarchical division information, thus forming a complete correspondence between task types and execution scopes. For example, the system obtains that the task type corresponding to the hydrological intelligent agent is watershed inflow prediction, and the execution scope is the processing of hydrological monitoring data for the entire watershed; the task type corresponding to the power generation intelligent agent is unit output optimization, and the execution scope is matching power generation constraints of each cascade power station; the task type corresponding to the flood control intelligent agent is reservoir safety management, and the execution scope is the control of water levels and outflow of each reservoir.

[0053] Step S32: Determine the task flow and result aggregation method between the intelligent agents according to the task type and the execution scope; It should be noted that "task type" refers to the category of scheduling business handled by each intelligent agent; "execution scope" refers to the business area and management level coverage that each intelligent agent can carry out business processing; "task flow" refers to the way scheduling tasks are passed, triggered, and continued between different intelligent agents; and "result aggregation" refers to the processing method in which the inference results output by each intelligent agent are collected, organized, and reported. In this embodiment, the system determines the task flow and result aggregation methods based on the task type and execution scope, enabling scheduling tasks to be passed in an orderly manner according to business logic, and allowing the output results of each intelligent agent to be collected uniformly, providing a data foundation for subsequent master intelligent agent fusion decision-making. In one possible implementation, the system uses a combination of sequential triggering and parallel processing to determine the task flow path.

[0054] Specifically, the system sets the initiation order, triggering conditions, and transmission paths of tasks among different agents according to the execution logic of scheduling operations. It also sets the reporting timing, aggregation channels, and processing rules for the output results of each agent, forming a standardized task flow and result aggregation method. For example, the system sets the hydrological agent to prioritize executing the water inflow prediction task and output the prediction results, then transfers the results to the power generation agent and the flood control agent to execute sub-tasks respectively, and finally aggregates the results of the three sub-tasks to the main agent for further processing.

[0055] Step S33: Based on the decision boundary, the data interaction permissions, and the task flow and result aggregation method, a multi-level intelligent agent collaborative workflow is formed.

[0056] It should be noted that the decision boundary refers to the business scope within which each agent can independently conduct reasoning and decision-making; data interaction permissions refer to the data types and data ranges that each agent can read, send, and invoke; task flow refers to the way scheduling tasks are passed, triggered, and continued between different agents; result aggregation refers to the processing method in which the reasoning results output by each agent are collected, organized, and reported; and multi-level agent collaborative workflow refers to the complete execution link in which multiple agents complete task processing, data interaction, and decision collaboration according to fixed rules. In this embodiment, the system forms a collaborative workflow based on the decision boundary, data interaction permissions, task flow, and result aggregation methods. This can integrate dispersed agent tasks into a unified and coherent scheduling and processing link, achieving standardized execution of the entire process of task decomposition, parallel processing, and result aggregation, thereby improving overall scheduling efficiency and stability. In one possible implementation, the system forms a complete closed-loop workflow including task initiation, agent execution, result aggregation, decision fusion, and instruction output.

[0057] Specifically, the system uses decision boundaries as the scope constraint for task execution, data interaction permissions as the compliance basis for data transmission, and task flow and result aggregation methods as the basis for execution sequence and data flow. These elements are combined and connected according to business logic to form a stable and repeatable multi-level intelligent agent collaborative workflow. For example, based on the decision boundaries, data interaction permissions, and set task flow and result aggregation methods of each intelligent agent, the system forms a multi-level intelligent agent collaborative workflow for the Wujiang cascade reservoirs: "task issuance—hydrological prediction—power generation optimization—flood control decision—result aggregation—fusion decision."

[0058] This embodiment further solidifies the collaboration logic and execution path of multi-level agents by clarifying the task types and execution scope of each agent, standardizing the task flow and result aggregation methods, and integrating them into a complete collaborative workflow. This effectively reduces the risk of agent collaboration conflicts and improves the orderliness, stability, and execution efficiency of watershed scheduling decisions.

[0059] In one feasible implementation, the step of providing scheduling and reasoning support for each intelligent agent based on a preset hydropower domain corpus, employing a dual mechanism of low-rank adaptation and retrieval enhancement, includes: Step S41: Obtain the watershed operation knowledge and scheduling procedures corresponding to the professional corpus in the hydropower field; It should be noted that the hydropower-related professional corpus refers to industry-specific data and text information used to support hydropower basin scheduling decisions; basin operation knowledge refers to professional knowledge reflecting the natural characteristics, hydrological patterns, and equipment operating status of a hydropower basin; and scheduling procedures refer to the management regulations and operational guidelines followed by cascade reservoirs within the basin for flood control and power generation scheduling. In this embodiment, the system acquires the basin operation knowledge and scheduling procedures corresponding to the hydropower-related professional corpus, extracting core decision-making basis that conforms to industry scenarios. This provides authentic, accurate, and compliant professional content for subsequent model adaptation and knowledge base construction, avoiding the problem of mismatch between general knowledge and industry scheduling needs. In one possible implementation, the system extracts the corresponding basin operation knowledge and scheduling procedures from historical operation data, standard specification documents, and scheduling management manuals.

[0060] Specifically, the system cleans, filters, and structures the acquired hydropower-related professional corpus, extracting basin operation knowledge and scheduling procedures that can be directly used for scheduling reasoning, forming standardized professional content that can be used for model training and knowledge base construction. For example, the system extracts basin operation knowledge such as the hydrological characteristics of cascade reservoirs, reservoir water level change patterns, and flood season safety constraints from the hydropower-related professional corpus of the Wujiang River Basin, as well as scheduling procedures such as power generation scheduling processes, flood control requirements, and equipment operation specifications.

[0061] Step S42: For any of the aforementioned intelligent agents, based on the knowledge of the watershed operation and the scheduling procedure, perform low-rank adaptation processing on the basic large model, construct a knowledge base based on the low-rank adaptation processing results and perform retrieval enhancement to form scheduling reasoning support capability.

[0062] It should be noted that the basic large model refers to a general-purpose large language model used for semantic understanding and reasoning computation; low-rank adaptation processing refers to a model optimization method that injects lightweight industry knowledge into the basic large model; the knowledge base refers to a dataset that stores professional knowledge in the hydropower field and supports fast retrieval; retrieval enhancement refers to a processing method that retrieves relevant knowledge from the knowledge base based on semantic matching to assist reasoning; and scheduling reasoning support capability refers to the ability to provide the agent with decision-making and result output capabilities that conform to industry rules and adapt to the actual situation of the watershed. In this embodiment, the system performs low-rank adaptation processing on the basic large model based on watershed operation knowledge and scheduling procedures, which can inject industry knowledge into the general model to improve scenario adaptability. Then, by building a knowledge base and performing retrieval enhancement, it can provide the agent with real-time and accurate knowledge support, making the reasoning results output by the agent more in line with the actual needs of watershed scheduling. In one possible implementation, the system uses a vector database to build a knowledge base and performs semantic similarity retrieval to achieve retrieval enhancement.

[0063] Specifically, the system uses basin operation knowledge and scheduling procedures as training content, performs low-rank adaptation processing on the basic large model to achieve industry-specific adaptation, then stores the processed professional knowledge in a knowledge base and configures retrieval strategies. When the intelligent agent performs inference, it calls the knowledge base to complete knowledge supplementation and verification, ultimately forming a stable and reliable scheduling inference support capability. For example, based on the Wujiang River basin operation knowledge and scheduling procedures, the system performs low-rank adaptation processing on the basic large model, constructs a Wujiang hydropower-specific knowledge base, and performs retrieval enhancement, providing scheduling inference support capabilities adapted to basin scenarios for hydrological intelligent agents, power generation intelligent agents, and flood control intelligent agents, respectively.

[0064] This embodiment improves the reasoning adaptability and accuracy of the agent in hydropower scheduling scenarios by extracting standardized watershed operation knowledge and scheduling procedures, performing low-rank adaptation processing on the basic large model, and building a knowledge base to enhance retrieval. It solves the technical problems of the general large model lacking industry knowledge and the reasoning results not matching actual scheduling needs.

[0065] In one feasible implementation, the acquisition of watershed hydrological monitoring data, power generation constraint data, and flood control safety data drives multi-level intelligent agents to perform collaborative reasoning, obtaining several target reasoning results. A main intelligent agent then performs a fusion decision on each of the target reasoning results to obtain the fusion decision result and the reasoning basis information corresponding to the fusion decision result, including: Step S51: Based on the hydrological monitoring data, drive the hydrological intelligent agent to generate hydrological prediction and inference results; It should be noted that hydrological monitoring data refers to data obtained from real-time monitoring of elements such as precipitation, flow, water level, and reservoir capacity within the basin. A hydrological intelligent agent refers to an intelligent agent specifically configured to handle hydrological-related calculations and predictions. Hydrological prediction inference results refer to the predicted trends and water volume of the basin in the future, derived from hydrological monitoring data. In this embodiment, the system drives the hydrological intelligent agent to perform inference calculations based on hydrological monitoring data. Leveraging industry-enhanced model capabilities and a professional knowledge base, it can obtain hydrological prediction conclusions that fit the characteristics of the basin, providing a data basis for subsequent power generation scheduling and flood control decisions. In one possible implementation, the hydrological intelligent agent completes the prediction and inference of future water inflow based on historical hydrological patterns and real-time monitoring data.

[0066] Specifically, the system inputs standardized hydrological monitoring data into a hydrological agent, which then utilizes scheduling and inference capabilities to perform trend analysis and numerical prediction on the data, ultimately generating hydrological prediction inference results that reflect future water inflow conditions in the basin. For example, the system inputs real-time precipitation, flow, and water level data from the Wujiang River basin into the hydrological agent, which then generates hydrological prediction inference results related to future water inflow forecasts for the basin.

[0067] Step S52: Based on the power generation constraint data, drive the power generation agent to generate power generation optimization inference results; It should be noted that power generation constraint data refers to relevant data that constrains the power generation process, such as unit operating limitations, output limits, power generation plans, and equipment status. The power generation agent refers to an agent specifically configured to handle power generation efficiency optimization and output allocation. The power generation optimization inference result refers to the optimized conclusions regarding unit output, power generation duration, and power generation revenue obtained under the constraint conditions. In this embodiment, the system drives the power generation agent to perform inference calculations based on the power generation constraint data, maximizing power generation efficiency while maintaining compliant operation, and providing a power generation-side optimization scheme for multi-objective balance. In one possible implementation, the power generation agent completes the optimal allocation inference of power generation output and duration based on an optimization algorithm.

[0068] Specifically, the system inputs power generation constraint data into the power generation agent, which then combines hydrological prediction and inference results with scheduling procedures to perform optimization calculations within the constraints, ultimately generating power generation optimization inference results that balance equipment safety and efficiency. For example, the system inputs power generation constraint data such as cascade power station unit output limits and maintenance plans into the power generation agent to generate power generation optimization inference results for the optimal unit power generation arrangement for the corresponding time period.

[0069] Step S53: Based on the flood control safety data, drive the flood control intelligent agent to generate flood control safety inference results; It should be noted that flood control safety data refers to relevant data ensuring flood control safety, such as reservoir warning water level, flood limit water level, upper limit of outflow, and safe storage capacity of the reservoir area. The flood control intelligent agent refers to an intelligent agent specifically configured to handle reservoir flood control safety and risk control. The flood control safety inference result refers to the conclusions on water level control and outflow control obtained under the premise of ensuring flood control safety. In this embodiment, the system drives the flood control intelligent agent to perform inference calculations based on flood control safety data, ensuring that reservoir operation meets flood control safety requirements, avoiding basin flood risks, and providing a safety-side constraint basis for multi-objective balance. In one possible implementation, the flood control intelligent agent completes risk control and flow control inference based on reservoir safety operation rules.

[0070] Specifically, the system inputs flood control safety data into the flood control intelligent agent, which then combines hydrological prediction and reasoning results with flood control regulations to conduct safety verification and control calculations, ultimately generating flood control safety reasoning results that ensure reservoir flood control safety. For example, the system inputs flood control safety data such as reservoir flood control limit water level and allowable discharge flow into the flood control intelligent agent, generating flood control safety reasoning results for the corresponding time period regarding reservoir water level and discharge control.

[0071] Step S54: Based on the hydrological prediction reasoning results, the power generation optimization reasoning results, and the flood control safety reasoning results, the main intelligent agent performs multi-objective fusion to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result.

[0072] It should be noted that the hydrological prediction inference result refers to the future water inflow prediction conclusion output by the hydrological intelligent agent; the power generation optimization inference result refers to the power generation benefit optimization conclusion output by the power generation intelligent agent; the flood control safety inference result refers to the flood control safety control conclusion output by the flood control intelligent agent; the main intelligent agent refers to the core intelligent agent used to coordinate the results of various sub-intelligent agents and execute multi-objective balance decisions; multi-objective fusion refers to the process of unifying and balancing multiple objectives such as flood control safety, power generation benefits, and hydrological constraints; the decision fusion result refers to the final scheduling scheme that takes into account multiple objective constraints; and the inference basis information refers to the rules, data, and calculation process descriptions that support the formation of the decision fusion result. In this embodiment, the main intelligent agent can resolve the objective conflict between flood control and power generation by performing multi-objective fusion on multiple types of inference results, obtain a directly executable scheduling scheme, and simultaneously output interpretable basis information, thereby improving the credibility and superviseability of the decision. In one possible implementation, the main intelligent agent performs balance fusion on multiple objectives according to a preset weight allocation strategy.

[0073] Specifically, the main intelligent agent receives hydrological prediction inference results, power generation optimization inference results, and flood control safety inference results. Based on the reservoir's operating status and constraints, it performs weight allocation and balance calculations on multiple objectives to form a decision fusion result that meets the actual scheduling needs of the basin. Simultaneously, it records and generates the inference basis information supporting this result. For example, based on the three types of inference results—hydrological prediction, power generation optimization, and flood control safety—in the Wujiang River basin, the main intelligent agent performs multi-objective balance fusion to obtain a unified scheduling scheme for the cascade reservoirs and generates the corresponding inference basis information for this scheduling scheme.

[0074] This embodiment completes hydrological prediction, power generation optimization, and flood control safety-specific reasoning through sub-agents, while the main agent executes multi-objective fusion decision-making, realizing the synergistic optimization of flood control safety and power generation benefits. It effectively solves the technical problems of a single agent being unable to take multiple objectives into account, decision conflicts, and insufficient accuracy.

[0075] Based on this, this application provides a method for water current domain scheduling, referring to... Figure 2 , Figure 2 This is a flowchart illustrating Embodiment 2 of the hydroelectric current domain scheduling method of this application.

[0076] In one feasible implementation, the step of performing multi-objective fusion through the main intelligent agent based on the hydrological prediction reasoning result, the power generation optimization reasoning result, and the flood control safety reasoning result to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result includes: Step S61: Obtain the current reservoir operating water level and reservoir capacity constraint information for the current watershed; It should be noted that the current watershed refers to the hydroelectric range of the cascade reservoirs where scheduling decisions are being implemented; the reservoir operating water level refers to the real-time water level data of the reservoir currently in operation; and the reservoir capacity constraint information refers to the upper limit, lower limit, and related constraints of the adjustable reservoir capacity allowed within the current scheduling cycle. In this embodiment, the system acquires the reservoir operating water level and capacity constraint information of the current watershed, which provides the most direct basis for reservoir status for multi-objective balance fusion, ensuring that subsequent decisions conform to the actual operating conditions of the reservoir and avoiding discrepancies between decisions and actual operating conditions. In one possible implementation, the system collects the reservoir operating water level and capacity constraint information from the real-time reservoir monitoring system.

[0077] Specifically, the system reads the current operating water level values ​​of reservoirs from real-time monitoring equipment and data platforms in the basin, and extracts the upper and lower limits of reservoir capacity and adjustable capacity constraints within the corresponding scheduling cycle to form reservoir operating water level and capacity constraint information for decision support. For example, the system obtains the current real-time operating water level data of the Wujiang cascade reservoirs, as well as flood season capacity constraints and daily scheduling capacity constraints.

[0078] Step S62: Based on the reservoir operating water level and the reservoir capacity constraint information, determine the basin inflow trend and water allocation constraints according to the hydrological prediction reasoning results, determine the unit power generation output and power generation duration allocation scheme according to the power generation optimization reasoning results, and determine the reservoir safe water level and outflow constraints according to the flood control safety reasoning results, so as to perform multi-objective balance fusion, generate the decision fusion result under the conventional scheduling scenario and the reasoning basis information corresponding to the decision fusion result.

[0079] It should be noted that the inflow trend of the basin refers to the direction and magnitude of changes in the water volume of the basin in the future; the water allocation constraint refers to the available water volume limit for each reservoir and each time period determined based on the inflow situation; the power generation output of the generating units refers to the planned power output value of the generating units; the power generation time allocation scheme refers to the power generation time arrangement of each unit within the scheduling cycle; the reservoir safe water level refers to the highest operating water level allowed to ensure flood control safety; the downstream flow constraint refers to the maximum downstream flow allowed to ensure downstream safety; the multi-objective balance fusion refers to the process of coordinating the three types of objectives: hydrological constraints, power generation benefits, and flood control safety; the routine scheduling scenario refers to the daily scheduling scenario under non-extreme weather and non-emergency conditions; the decision fusion result refers to the final scheduling scheme formed by taking into account multiple constraints and objectives; and the reasoning basis information refers to the data sources, rule basis, and calculation process that support the formation of the decision fusion result.

[0080] In this embodiment, the system uses reservoir operating water level and capacity constraints as a basis to unify and balance the three types of reasoning results: hydrology, power generation, and flood control. This enables the generation of an optimal scheduling plan under safe, compliant, and feasible conditions, while simultaneously forming interpretable and traceable reasoning information, thus improving the scientific rigor and transparency of decision-making. In one possible implementation, the system performs multi-objective balance fusion according to the principles of safety priority and optimal efficiency.

[0081] Specifically, the system uses reservoir operating water level and capacity constraints as basic decision-making conditions. Based on hydrological forecasting and reasoning results, it analyzes and determines future inflow trends and available water allocation constraints. Based on power generation optimization and reasoning results, it determines the allocation scheme for unit power output and power generation duration. Based on flood control safety and reasoning results, it determines the reservoir safety water level and discharge flow constraints. Under the condition that all three types of constraints are satisfied simultaneously, it performs multi-objective balance calculations to generate decision fusion results for routine scheduling scenarios and simultaneously compiles the corresponding reasoning basis information. For example, based on the current operating water level and capacity constraints of the Wujiang cascade reservoirs, combined with hydrological forecasts of inflow trends, power generation optimization schemes, and flood control safety water level limits, the system performs multi-objective balance fusion to generate comprehensive decision fusion results for reservoir water release, power generation, and water storage under daily scheduling, and provides the corresponding reasoning basis information.

[0082] In another embodiment, the system acquires the current reservoir operating water level and storage capacity constraint information of the basin, monitors the issuance of extreme rainfall warning information in the basin, and uses the reservoir operating water level and storage capacity constraint information as the basis for decision-making. Based on the hydrological prediction reasoning results, the system determines the future heavy rainfall trend and excess water allocation constraints of the basin. Based on the power generation optimization reasoning results, the system dynamically reduces the power generation output of the units and shortens the power generation duration allocation scheme. Based on the flood control safety reasoning results, the system strictly controls the reservoir safety water level and limits the upper limit of the discharge flow. The system performs multi-objective balance fusion with flood control safety as the priority, and generates the decision fusion result under extreme weather scenarios and the reasoning basis information corresponding to the decision fusion result.

[0083] This embodiment uses real-time reservoir water level and capacity constraints as the basis for decision-making, and integrates multi-dimensional reasoning results from hydrology, power generation, and flood control to achieve multi-objective balance and fusion. Under conventional scheduling scenarios, it generates safe, feasible, and optimal scheduling decisions, while forming interpretable supporting information. This effectively solves the technical problems of conflict between flood control and power generation objectives, lack of decision-making basis, and insufficient accuracy.

[0084] In one feasible implementation, the construction of a penetrating cognitive decision-making display interface, which pushes the decision fusion result and the corresponding reasoning basis information to management personnel, realizes cross-level penetrating display of decision information and issuance of instructions, includes: Step S71: Obtain the display elements and risk warnings corresponding to the decision fusion result and the reasoning basis information; It should be noted that the decision fusion result refers to the final scheduling scheme generated by the main intelligent agent that balances flood control safety and power generation benefits; the reasoning basis information refers to the data, rules, and calculation processes supporting the formation of the decision fusion result; the display elements refer to the decision content, parameters, curves, and status information presented in the interface; and the risk warnings refer to potential operational risks identified based on the scheduling scheme, such as water level exceeding limits, insufficient output, and excessive discharge. In this embodiment, the system acquires the display elements and risk warnings corresponding to the decision fusion result and the reasoning basis information, which can transform abstract scheduling conclusions into intuitively displayable and quickly identifiable interface content, providing managers with clear, complete, and superviseable decision information. In one possible implementation, the system uses water level data, output data, discharge data, and constraint data as unified display elements.

[0085] Specifically, the system extracts key parameters and execution content related to the scheduling scheme from the decision fusion results, and extracts publicly displayable rules and calculation processes from the reasoning basis information to form standardized display elements. Simultaneously, it identifies risks to the decision scheme based on operational constraints and generates corresponding risk warning information. For example, the system extracts display elements such as water level, downstream flow, and unit output from the cascade reservoir scheduling decision fusion results and reasoning basis information, and generates a risk warning that the water level is approaching the control threshold.

[0086] Step S72: Based on the displayed elements and the risk warning, construct a penetrating cognitive decision display interface, wherein the decision fusion result is simulated and the effect is pre-done through a watershed digital twin model, and synchronized to the penetrating cognitive decision display interface; It should be noted that the "display elements" refer to the information items used to present the content and basis of the decision-making process; the "risk warning" refers to the potential operational risk information in the scheduling plan; the "penetrating cognitive decision-making display interface" refers to a visual operation interface that supports cross-level data drill-down, makes the decision-making process visible, and makes the reasoning basis verifiable; the "basin digital twin model" refers to a digital mapping model that reflects the real state of the basin's reservoirs, generating units, and river channels; the "simulation and deduction" refers to the processing of simulating the execution process of the scheduling plan; and the "effect pre-play" refers to the processing of displaying the water level, flow rate, and power generation status after the execution of the scheduling plan in advance. In this embodiment, the system constructs a penetrating cognitive decision-making display interface based on the display elements and risk warnings, and combines it with digital twins for simulation and pre-play, which can achieve interpretability, verifiability, and traceability of the entire decision-making process, solving the problem that traditional interfaces only display results and not processes. In one possible implementation, the system uses a combination of visual charts and twin scenes to construct the display interface.

[0087] Specifically, the system arranges the displayed elements and risk warnings according to a hierarchical structure, forming a transparent display interface that can be viewed from the group level down to the plant level. Simultaneously, it uses a digital twin model of the river basin to simulate and pre-demonstrate the results of the decision-making process, and outputs the simulation data and pre-demonstration effects to the display interface. For example, the system constructs a transparent cognitive decision-making display interface based on displayed elements such as water level, discharge, and output, along with risk warnings. It then uses the digital twin model of the Wujiang River basin to simulate and extrapolate the scheduling plan, displaying the pre-demonstration effects synchronously on the interface.

[0088] Step S73: Based on the penetrating cognitive decision display interface, push decision information to management personnel and complete cross-level instruction issuance and manual intervention.

[0089] It should be noted that the "penetrating cognitive decision-making display interface" refers to an operation interface that supports cross-level viewing, manual adjustment, and instruction issuance. "Decision information" refers to the overall content of decision fusion results, reasoning basis information, risk warnings, and simulation effects. "Management personnel" refers to staff responsible for scheduling approval and supervision at the group, watershed, and plant levels. "Cross-level instruction issuance" refers to the process of pushing scheduling instructions from the management level to the execution terminal. "Manual intervention" refers to the operations of management personnel in adjusting, modifying, confirming, and rejecting scheduling plans. In this embodiment, the system pushes decision information to management personnel based on the display interface, achieving decision transparency, direct instruction delivery to the site, and support for real-time adjustments, thereby improving the execution efficiency and control of scheduling instructions. In one possible implementation, the system synchronously completes the push of decision information and the issuance of instructions through a mobile terminal and the platform.

[0090] Specifically, the system pushes complete decision information to managers with corresponding permissions through a penetrating cognitive decision-making display interface. After viewing, managers can confirm or adjust the information. The system then pushes the final confirmed dispatch instructions across levels to the on-site execution units, while retaining records of manual intervention and instruction execution. For example, the system pushes dispatch decision information for the Wujiang cascade reservoirs to managers through the penetrating cognitive decision-making display interface. After confirmation by the managers, the system pushes the dispatch instructions across levels to the power station site and supports managers in real-time adjustments to the plan.

[0091] This embodiment constructs a decision display interface that is demonstrable, simulating, penetrable, and interventionable, enabling the entire scheduling decision-making process to be explainable, supervised, and executable. It effectively solves the technical problems of cross-level information transmission deviation, lack of decision-making basis, and untimely issuance of on-site instructions.

[0092] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0093] This application also provides a hydroelectric domain scheduling device, please refer to... Figure 3 The hydroelectric current domain scheduling device includes: The acquisition module 31 is used to acquire the business scope and hierarchical management information of hydropower domain scheduling, construct a multi-level intelligent agent collaborative architecture based on access control protocol, and determine the decision boundary and interaction rules of each intelligent agent. Among them, according to the preset hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is adopted to provide scheduling reasoning support for each intelligent agent. Decision module 32 is used to acquire watershed hydrological monitoring data, power generation constraint data and flood control safety data, drive multi-level intelligent agents to perform collaborative reasoning, obtain several target reasoning results, and perform fusion decision on each of the target reasoning results through the main intelligent agent to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result; The push module 33 is used to construct a penetrating cognitive decision display interface, push the decision fusion result and the reasoning basis information corresponding to the decision fusion result to the management personnel, and realize the cross-level penetrating display of decision information and the issuance of instructions.

[0094] The hydroelectric current domain scheduling device provided in this application, employing the hydroelectric current domain scheduling method described in the above embodiments, can solve the technical problems in the background art. Compared with the prior art, the beneficial effects of the hydroelectric current domain scheduling device provided in this application are the same as those of the hydroelectric current domain scheduling method described in the above embodiments, and other technical features in the hydroelectric current domain scheduling device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0095] This application provides a water current domain scheduling device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the water current domain scheduling method in the above embodiment 1.

[0096] The following is for reference. Figure 4 The diagram illustrates a structural schematic of a hydroelectric current domain scheduling device suitable for implementing embodiments of this application. The hydroelectric current domain scheduling device in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 4 The hydroelectric current domain scheduling device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0097] like Figure 4As shown, the hydroelectric current domain scheduling device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the hydroelectric current domain scheduling device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the hydroelectric current domain scheduling equipment to communicate wirelessly or wiredly with other equipment to exchange data. Although the figure shows hydroelectric current domain scheduling equipment with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0098] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0099] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations 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. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0100] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the hydroelectric domain scheduling methods provided by the above methods.

[0101] The system embodiments described above are merely illustrative. 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; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0102] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0103] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A hydroelectric current domain scheduling method, characterized in that, include: The system acquires the scope and hierarchical management information of hydropower domain scheduling business, constructs a multi-level intelligent agent collaborative architecture based on access control protocol, and determines the decision boundaries and interaction rules of each intelligent agent. In particular, based on the preset hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is adopted to provide scheduling reasoning support for each intelligent agent. The system acquires watershed hydrological monitoring data, power generation constraint data, and flood control safety data, drives multi-level intelligent agents to perform collaborative reasoning, obtains several target reasoning results, and then uses the main intelligent agent to perform fusion decision-making on each of the target reasoning results to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result. A penetrating cognitive decision-making display interface is constructed, which pushes the decision fusion result and the corresponding reasoning basis information to the management personnel, so as to realize the cross-level penetrating display of decision information and the issuance of instructions.

2. The hydroelectric current domain scheduling method as described in claim 1, characterized in that, The process of acquiring the scope and hierarchical management information of hydropower domain scheduling services, constructing a multi-level intelligent agent collaborative architecture based on access control protocols, and determining the decision boundaries and interaction rules of each intelligent agent includes: Based on the scope of the hydroelectric current domain scheduling business and the hierarchical management information, the hierarchical division information is determined; Based on the hierarchical division information and the access control protocol, the decision boundaries and data interaction permissions of each intelligent agent are determined; Based on the decision boundaries and data interaction permissions, establish interaction rules and collaboration mechanisms for multi-level intelligent agents.

3. The hydroelectric current domain scheduling method as described in claim 2, characterized in that, The step of establishing multi-level agent interaction rules and collaboration mechanisms based on the decision boundary and the data interaction permissions includes: Obtain the task type and execution scope corresponding to each of the aforementioned intelligent agents; Based on the task type and the execution scope, determine the task flow and result aggregation method among the intelligent agents; Based on the decision boundaries, data interaction permissions, and task flow and result aggregation methods, a multi-level intelligent agent collaborative workflow is formed.

4. The hydroelectric current domain scheduling method as described in claim 1, characterized in that, The method, based on a pre-set hydropower domain corpus, employs a dual mechanism of low-rank adaptation and retrieval enhancement to provide scheduling and reasoning support for each intelligent agent, including: Acquire the watershed operation knowledge and scheduling procedures corresponding to the professional corpus in the hydropower field; For any of the aforementioned intelligent agents, based on the knowledge of the watershed operation and the scheduling procedure, a low-rank adaptation process is performed on the basic large model. A knowledge base is constructed based on the results of the low-rank adaptation process, and retrieval enhancement is performed to form a scheduling reasoning support capability.

5. The hydroelectric current domain scheduling method as described in claim 1, characterized in that, The process involves acquiring watershed hydrological monitoring data, power generation constraint data, and flood control safety data, driving multi-level intelligent agents to perform collaborative reasoning, obtaining several target reasoning results, and then using a main intelligent agent to perform fusion decision-making on each of the target reasoning results to obtain the fusion decision result and the reasoning basis information corresponding to the fusion decision result, including: Based on the hydrological monitoring data, the hydrological intelligent agent is driven to generate hydrological prediction and inference results. Based on the power generation constraint data, the power generation agent is driven to generate power generation optimization inference results. Based on the flood control safety data, the flood control intelligent agent is driven to generate flood control safety inference results; Based on the hydrological prediction reasoning results, the power generation optimization reasoning results, and the flood control safety reasoning results, the main intelligent agent performs multi-objective fusion to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result.

6. The hydroelectric current domain scheduling method as described in claim 5, characterized in that, The process involves the main intelligent agent performing multi-objective fusion based on the hydrological prediction inference results, the power generation optimization inference results, and the flood control safety inference results to obtain the decision fusion result and the inference basis information corresponding to the decision fusion result, including: Obtain information on the current reservoir operating water level and capacity constraints in the watershed; Based on the reservoir operating water level and the reservoir capacity constraint information, the basin inflow trend and water allocation constraints are determined according to the hydrological prediction reasoning results; the unit power generation output and power generation duration allocation scheme is determined according to the power generation optimization reasoning results; and the reservoir safe water level and discharge flow constraints are determined according to the flood control safety reasoning results. Multi-objective balance fusion is performed to generate decision fusion results under conventional scheduling scenarios and the reasoning basis information corresponding to the decision fusion results.

7. The hydroelectric current domain scheduling method as described in claim 1, characterized in that, The construction of a penetrating cognitive decision-making display interface pushes the decision fusion result and the corresponding reasoning basis information to management personnel, realizing cross-level penetrating display of decision information and instruction issuance, including: Obtain the display elements and risk warnings corresponding to the decision fusion results and the reasoning basis information; Based on the displayed elements and the risk warnings, a penetrating cognitive decision-making display interface is constructed. The decision fusion results are simulated and the effects are previewed using a watershed digital twin model, and then synchronized to the penetrating cognitive decision-making display interface. Based on the penetrating cognitive decision display interface, decision information is pushed to management personnel, and cross-level instructions and manual intervention are completed.

8. A hydroelectric current domain scheduling device, characterized in that, include: The acquisition module is used to acquire the business scope and hierarchical management information of hydropower domain scheduling, construct a multi-level intelligent agent collaborative architecture based on access control protocol, and determine the decision boundary and interaction rules of each intelligent agent. Among them, based on the preset hydropower domain professional corpus, a dual mechanism of low-rank adaptation and retrieval enhancement is adopted to provide scheduling reasoning support for each intelligent agent. The decision module is used to acquire watershed hydrological monitoring data, power generation constraint data and flood control safety data, drive multi-level intelligent agents to perform collaborative reasoning, obtain several target reasoning results, and perform fusion decision on each of the target reasoning results through the main intelligent agent to obtain the decision fusion result and the reasoning basis information corresponding to the decision fusion result. The push module is used to build a penetrating cognitive decision display interface, push the decision fusion result and the reasoning basis information corresponding to the decision fusion result to the management personnel, and realize the cross-level penetrating display of decision information and the issuance of instructions.

9. A hydroelectric current domain scheduling device, characterized in that, The hydroelectric current domain scheduling device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the hydroelectric current domain scheduling method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the hydroelectric domain scheduling method as described in any one of claims 1 to 7.

Images