Thermal power plant operation management method and system based on multiple agents

Abstract

The invention provides a multi-agent-based thermal power plant operation management method and system, and belongs to the technical field of agents, and the method specifically comprises the steps: taking adaptive data of agents of thermal equipment and an adjustment instruction and adaptive data of other thermal equipment in a combined adjustment process as a basis; determining response deviation types of the agents of the thermal equipment, and when the joint adjustment data of the agents of different response deviation types meet requirements, determining the joint adjustment data of the agents of different response deviation types after the agents of the target updating processing strategy are updated based on the joint adjustment data of the agents and the agent data of the target updating processing strategy, and determining the joint adjustment data of the agents of the target updating processing strategy after the agents of the target updating processing strategy are updated. The data interaction strategy of the intelligent agent improves the reliability of operation management processing of the thermal power plant.

Description

Technical Field

[0001] This invention belongs to the field of intelligent agent technology, and in particular relates to a method and system for the operation and management of thermal power plants based on multiple intelligent agents. Background Technology

[0002] Thermal power plants have a wide variety of equipment types. Existing technical solutions often use a unified monitoring platform to monitor and process the equipment types. However, due to the large number of devices, existing technical solutions may not be able to effectively achieve comprehensive monitoring and processing of the various thermal equipment in a thermal power plant.

[0003] To address the aforementioned technical issues, existing solutions utilize intelligent agents to monitor, analyze, and process equipment. However, due to differences in the joint regulation data of different thermal equipment responding to regulation commands, determining the data interaction strategy of the intelligent agent after updating the data based on the joint regulation data is crucial to ensuring that the updated intelligent agent can meet regulation requirements and thus guarantee the operational stability of the thermal power plant.

[0004] Specifically, this application provides a multi-agent-based method and system for the operation and management of thermal power plants. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a multi-agent-based method for the operation and management of thermal power plants, which includes: S1 uses the operating data of the thermal power plant as a basis to determine the adjustment data of different thermal equipment in the thermal power plant. Based on the adjustment data and the joint adjustment data with thermal equipment without intelligent agents, the S1 determines the update processing strategy of the intelligent agent of the thermal equipment. When the update processing strategy does not belong to the target update processing strategy, the S1 proceeds to the next step. S2 determines the response deviation type of the intelligent agent of the thermal equipment based on the adaptation data of the intelligent agent of the thermal equipment and the adjustment command, as well as the adaptation data of other thermal equipment in the joint adjustment process. When the joint adjustment data of intelligent agents with different response deviation types meet the requirements, proceed to the next step. S3 determines the data interaction strategy of the agents after the agent updates the target update processing strategy, based on the joint adjustment data between different agents and the agent data of the target update processing strategy.

[0006] The beneficial effects of this invention are as follows: Based on regulation data and joint regulation data with thermal equipment without intelligent agents, an update processing strategy for the intelligent agents of thermal equipment is determined. This takes into account the frequency of regulation processing by the intelligent agents of thermal equipment and the difference in reliability of joint regulation processing caused by the difference in the frequency of joint regulation processing with thermal equipment without intelligent agents. Thus, the update strategy is determined from the perspective of the reliability of joint regulation processing, thereby ensuring the operational stability and reliability of the thermal power plant.

[0007] Based on the joint regulation data between different agents and the agent data of the target update processing strategy, the data interaction strategy of the agents after the target update processing strategy is updated is determined. This not only takes into account the frequency of joint regulation processing between the agents and the agents of the target update processing strategy, but also the number of agents of the target update processing strategy. Thus, based on the requirements of joint regulation processing and the frequency of update processing, the data interaction strategy is determined to simulate whether the updated agents can meet the requirements of the regulation instructions by simulating data interaction among multiple agents after the update. This ensures the stability and security of the updated agents while reducing the impact on the normal operation of the agents.

[0008] Furthermore, the thermal equipment includes a coal mill, a coal feeder, a fan, a water pump, a boiler, a steam turbine, and a desuperheater / pressure reducer.

[0009] Furthermore, the adjustment data of the thermal equipment includes the number of times the thermal equipment has undergone adjustment processing in history.

[0010] Furthermore, the method for determining the update processing strategy of the intelligent agent of the thermal equipment is as follows: Based on the aforementioned adjustment data, the process of adjustment processing of the thermal equipment in history is determined and used as the adjustment processing process. Based on the joint regulation data of thermal equipment without intelligent agents during the regulation process, identify the thermal equipment without intelligent agents that are subject to joint regulation during the regulation process, and designate them as thermal equipment with regulation defects. Based on the data of the thermal equipment with regulatory defects in different regulation processes, an update processing strategy for the intelligent agent of the thermal equipment is determined.

[0011] Furthermore, the method for determining the data interaction strategy of the intelligent agent is as follows: The agent that adopts the target update processing strategy is taken as the target policy agent. The joint adjustment data of the target policy agent and the agent is used to determine the adjustment process of the joint adjustment processing of the target policy agent and the agent. Based on the agent data of the target update processing strategy, determine the number of agents for the target update processing strategy; The data interaction strategy of the agent is determined based on the updated data of the target policy agent, the number of agents in the target update processing strategy, and the adjustment process of the target policy agent and the agent performing joint adjustment processing.

[0012] In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the aforementioned multi-agent-based thermal power plant operation management method when running the computer program.

[0013] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0014] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0015] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0016] Figure 1 This is a flowchart of a multi-agent-based operation and management method for thermal power plants; Figure 2 This is a flowchart illustrating the method for determining the update and processing strategy of the intelligent agent in thermal equipment. Figure 3 This is a flowchart of a method for determining the response deviation type of an intelligent agent in a thermal device. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0018] Example 1 like Figure 1 As shown, this application provides a multi-agent-based operation and management method for thermal power plants, specifically including: S1 uses the operating data of the thermal power plant as a basis to determine the adjustment data of different thermal equipment in the thermal power plant. Based on the adjustment data and the joint adjustment data with thermal equipment without intelligent agents, the S1 determines the update processing strategy of the intelligent agent of the thermal equipment. When the update processing strategy does not belong to the target update processing strategy, the S1 proceeds to the next step. S2 determines the response deviation type of the intelligent agent of the thermal equipment based on the adaptation data of the intelligent agent of the thermal equipment and the adjustment command, as well as the adaptation data of other thermal equipment in the joint adjustment process. When the joint adjustment data of intelligent agents with different response deviation types meet the requirements, proceed to the next step. S3 determines the data interaction strategy of the agents after the agent updates the target update processing strategy, based on the joint adjustment data between different agents and the agent data of the target update processing strategy.

[0019] Furthermore, the thermal equipment includes a coal mill, a coal feeder, a fan, a water pump, a boiler, a steam turbine, and a desuperheater / pressure reducer.

[0020] Furthermore, the adjustment data of the thermal equipment includes the number of times the thermal equipment has undergone adjustment processing in history.

[0021] Specifically, such as Figure 2 As shown, the method for determining the update processing strategy of the intelligent agent of the thermal equipment is as follows: Based on the aforementioned adjustment data, the process of adjustment processing of the thermal equipment in history is determined and used as the adjustment processing process. Based on the joint regulation data of thermal equipment without intelligent agents during the regulation process, identify the thermal equipment without intelligent agents that are subject to joint regulation during the regulation process, and designate them as thermal equipment with regulation defects. Based on the data of the thermal equipment with regulatory defects in different regulation processes, an update processing strategy for the intelligent agent of the thermal equipment is determined.

[0022] It is understandable that, based on the data of the thermal equipment with regulatory defects during different regulation processes, the update processing strategy of the intelligent agent of the thermal equipment is determined, specifically including: Based on the data of thermal equipment with regulatory defects in different regulation processes, the number of thermal equipment with regulatory defects in different regulation processes is determined. The process of adjusting the number of defective thermal equipment that does not meet the requirements is regarded as a risk adjustment process. Based on the risk adjustment process data, an update processing strategy for the intelligent agent of the thermal equipment is determined.

[0023] It is understood that the risk adjustment process is a process of adjusting the number of defective thermal devices to be greater than a preset threshold for the number of defective thermal devices.

[0024] It should be noted that when the number of risk adjustment processes of the thermal equipment does not meet the requirements, for example, when the average daily number of risk adjustment processes on different dates is greater than the preset number threshold, the update processing strategy of the intelligent agent of the thermal equipment is determined to be the target update processing strategy because the reliability of the adjustment process is poor. That is, when the update number of its adjustment process meets the requirements, the update processing of the intelligent agent of the thermal equipment is performed.

[0025] In one possible embodiment, when the number of updates in the adjustment process meets the requirements, the intelligent agent of the thermal device is updated, specifically including: If the number of new additions in the adjustment process since the last update of the agent exceeds a preset threshold, then the update process of the agent of the thermal equipment is performed.

[0026] It should also be noted that when the number of risk adjustment processes of the thermal equipment meets the requirements, the update processing strategy of the intelligent agent of the thermal equipment is determined to be not a target update processing strategy. That is, when the number of updates of its risk adjustment processes meets the requirements, the update processing of the intelligent agent of the thermal equipment is performed.

[0027] In one possible embodiment, if the number of new risk adjustment processes since the last update of the agent is greater than a preset threshold for the number of adjustment processes, then the update of the agent of the thermal equipment is performed.

[0028] Specifically, in a thermal power plant, the focus is on a particular main burner of boiler #1. This equipment is equipped with an agent to optimize its fuel and air ratio. This embodiment aims to demonstrate how to determine the timing of agent updates specifically for this one unit, based on its associated historical regulation data.

[0029] 2. Key Concepts and Parameter Presets Core Concepts (Regulating the Main Burner of Boiler #1): Regulation Process: Any in-plant regulation event requiring or affecting the main burner of boiler #1. Examples include load changes, fuel type switching, and adjustments to environmental indicators (such as NOx). Defective Thermal Equipment: During a regulation process, equipment that requires "joint regulation" with the intelligent system of the main burner of boiler #1, but lacks its own intelligent system (or its intelligent system fails to cooperate effectively), resulting in poor joint regulation. For example, when a load increase is needed, the intelligent system of boiler #1 instructs to increase fuel, but the secondary air fan, lacking an intelligent system, responds slowly or inaccurately, leading to an imbalance in the air-fuel ratio and fluctuations in the entire regulation process. In this case, the secondary air fan is recorded as a "defective thermal equipment" affecting the intelligent system of boiler #1 during this process.

[0030] Risk adjustment process: If too many "adjustment defect thermal devices" appear during the adjustment process of this device, it indicates that the collaborative environment of this agent is very poor during the adjustment and the reliability of its adjustment effect is not good.

[0031] Preset parameters (for this single device): Preset threshold for the number of defective thermal devices: 1 (In a single adjustment for this device, if one or more non-intelligent devices with poor coordination occur, the process is considered a risky process), Preset threshold for the number of risky processes per day: 0.1 (If the number of risky processes per day is greater than this value, it indicates that the daily operating environment of the agent is harsh, and the "target update processing strategy" needs to be activated), Preset threshold for the number of new processes (normal processes): 15 (used as a trigger condition under the target update strategy), Preset threshold for the number of adjustment processes (risky processes): 10 (used as a trigger condition under non-target update strategies); 3. Data Preparation and Calculation Process Step 1: Identify the “risk regulation process” for this equipment. The system analyzed all regulation processes involving the main burner of boiler #1 over the past 30 days and identified the “regulation defective thermal equipment” that did not coordinate well with it in each process.

[0032] Date D1: A small load increase process, all cooperating equipment (coal feeder, blower, etc.) responded well. Defective equipment = 0 -> Non-risk process. Date D2: A fuel calorific value fluctuation adjustment process, the secondary blower response overshooted, causing furnace pressure fluctuations. Defective equipment = 1 -> Risk process (1 >= 1). Date D3: A NOx reduction operation, the induced draft fan and burnout damper actions were not coordinated. Defective equipment = 2 -> Risk process. Dates D4 to D30: ... (Assuming that 2 risk adjustment processes targeting this equipment were identified in the remaining 27 days), statistical results (only for the main burner of boiler #1): In the past 30 days, the total number of risk adjustment processes = 1 (D2) + 1 (D3) + 2 = 4 times. In the past 30 days, the average number of risk adjustment processes per day = 4 / 30 ≈ 0.13 times / day.

[0033] Step 2: Determine the update strategy (target update processing strategy) for this device. Check if the average daily number of risky processes (0.13) exceeds the preset threshold (0.1). If the average daily number of risky processes does not meet the requirement (it exceeds the threshold), then the "target update processing strategy" is activated for this device. This indicates that the reliability of the collaborative working environment in which this agent operates remains poor, and an update needs to be prioritized.

[0034] Step 3: Under "Target Update Strategy", determine whether to trigger an update immediately. Since the target update strategy has been activated, the system uses the corresponding conditions to make judgments: "If the number of new additions in the adjustment process since the last update of the agent is greater than the preset number of new additions threshold, then an update will be performed."

[0035] Background information: The last update time of the intelligent agent of the main burner of boiler #1 was 50 days ago. Check the data: Since the last update, how many adjustment processes (including risk and non-risk) have occurred in total? Assuming the statistical result is 18 times, the judgment is: The number of new adjustment processes (18) > the preset new number threshold (15). Analysis: 18 > 15, the condition is met, the system determines that the intelligent agent of the main burner of boiler #1 should be updated immediately.

[0036] Operating Environment Health Assessment: The system first assesses the daily operating environment of this specific agent. It identifies an average of approximately 0.13 risk events per day, exceeding the tolerance threshold of 0.1. This means the agent is frequently surrounded by "unintelligent teammates" (devices that are not intelligent or poorly cooperating), making it difficult to achieve optimal performance. Therefore, the system marks it as a high-priority update target (target update processing strategy).

[0037] Update Timing: After being identified as high priority, the system checks its activity frequency. Since the last update, it has participated in 18 adjustment processes, exceeding the threshold of 15. This indicates that the agent has accumulated enough new operational experience (including a large amount of unsuccessful experience collaborating with "incompetent teammates"), providing a data foundation for model updates.

[0038] This embodiment precisely demonstrates how to "tailor-make" update strategies for each intelligent device within a thermal power plant. This method focuses not only on the device itself but also on its collaborative ecosystem with surrounding devices. By monitoring the number of "deficiently regulated thermal devices," the system can accurately identify which intelligent agents are operating in unfavorable environments and prioritize their updates and optimizations, thereby improving the overall collaborative efficiency and stability of the power plant operation.

[0039] Specifically, such as Figure 3 As shown, the method for determining the response deviation type of the intelligent agent of the thermal device is as follows: Based on the adaptation data of the intelligent agent and the adjustment command of the heating equipment, the adaptation status of the adjustment requirements of the intelligent agent and the adjustment command of the heating equipment is determined. Based on the adaptation situation, determine the adjustment process where the adjustment result of the intelligent agent of the thermal equipment does not match the adjustment requirements of the adjustment command, and treat it as an incompatible adjustment process; Based on the mismatch adjustment process and the matching status of other thermal devices with the adjustment command during the adjustment process, the response deviation type of the intelligent agent of the thermal device is determined.

[0040] It is understandable that when the intelligent agent of the thermal device does not have a mismatched adjustment process, the response deviation type of the intelligent agent of the thermal device is determined to be a matched response type.

[0041] Furthermore, if the intelligent agent of the heating device has a mismatched adjustment process, it is determined whether the number of mismatched adjustment processes of the intelligent agent of the heating device is greater than a preset threshold for the number of mismatched processes. If so, the response deviation type of the intelligent agent of the heating device is determined to be a mismatched abnormality type. If not, the response deviation type of the intelligent agent of the heating device is determined by using the matching situation of other heating devices and the adjustment requirements of the adjustment command during the adjustment process.

[0042] Specifically, by utilizing the matching situation between other thermal devices and the adjustment requirements of the adjustment command during the adjustment process, the response deviation type of the intelligent agent of the thermal device is determined, including: Based on the matching status of other thermal equipment with the adjustment requirements of the adjustment command in different adjustment processes, the adjustment process of other thermal equipment that does not match the adjustment requirements of the adjustment command is identified, and it is regarded as the joint adjustment deviation process. Based on the number of other thermal equipment that do not match the adjustment requirements of the adjustment command during the joint adjustment deviation process, the adjustment anomaly impact factor of the joint adjustment deviation process is determined. Based on the abnormal influence factors of regulation in different joint regulation deviation processes and the number of joint regulation deviation processes of the thermal equipment, the abnormal influence coefficient of the intelligent agent of the thermal equipment is determined. If the abnormal influence coefficient is greater than the preset influence coefficient threshold, then the response deviation type of the intelligent agent of the thermal equipment is determined to be a matching abnormal type; otherwise, the response deviation type of the intelligent agent of the thermal equipment is determined to be a matching response type.

[0043] In this embodiment, we focus on the intelligent agent of the coal feeder for boiler #2 in the thermal power plant. This agent is responsible for precisely controlling the coal feed rate according to load commands. Following prior process evaluation, the system has determined that its update processing strategy does not belong to the target update processing strategy. Now, a detailed diagnosis of the response performance of the coal feeder agent itself is required.

[0044] 2. Key Concepts and Parameter Presets Core concepts: Regulation instructions: such as "increase fuel supply to boost 5MW load", regulation requirements: the desired precise value of coal supply and the corresponding load boosting effect.

[0045] Mismatched adjustment process: After the process is completed, the actual coal feed or final load does not meet the adjustment requirements. Joint adjustment deviation process: In one adjustment, in addition to the #2 coal feeder, other equipment (such as blowers and induced draft fans) also have control deviations.

[0046] Adjusting the abnormal impact factor: Quantifying the degree of interference of other equipment deviations on the control of the coal feeder; Abnormal impact coefficient: Comprehensively assessing the impact of system interference on the coal feeder intelligent agent; Preset threshold for the number of mismatched processes: 3; Preset threshold for the impact coefficient: 0.7. 3. Data Preparation and Calculation Process The system analyzed 60 recent adjustment records of the coal feeder's intelligent agent.

[0047] Step 1: Preliminary screening and judgment, statistical analysis of mismatched processes: It was found that in these 60 processes, the control parameters of the #2 coal feeder did not meet the adjustment requirements after the process ended in 2 instances. These 2 processes were marked as mismatched adjustment processes.

[0048] First-level judgment: The number of mismatched processes (2) is less than the preset threshold for the number of mismatched processes (3). Therefore, it cannot be directly determined as a mismatch type and further analysis is required.

[0049] Step Two: In-depth analysis of joint regulation Identifying the Joint Regulation Deviation Process: A thorough analysis of these two mismatch processes revealed the following: Process A (Load Increase): The #2 coal feeder failed to reach the target coal feed rate. Simultaneously, the blower also failed to reach the target airflow, resulting in insufficient total airflow. Process B (Fuel Stabilization Adjustment): The #2 coal feeder's coal feed rate fluctuated excessively. Simultaneously, the primary blower pressure fluctuated, and the furnace temperature sensing element reported abnormal data. Conclusion: Both mismatch processes were joint regulation deviation processes.

[0050] Step 3: Calculate the moderating factor for abnormal influences Definition: Adjustment anomaly impact factor = Number of other mismatched devices in the process / Total number of devices participating in joint adjustment in the process. For process A: Number of other mismatched devices: 1 (forced draft fan); Number of main devices participating in joint adjustment: 4 (coal feeder, forced draft fan, induced draft fan, secondary air fan); Impact factor for process A = 1 / 4 = 0.25 Process B calculation: Number of other mismatched equipment: 2 (primary air fan, furnace temperature measuring element), Number of main equipment involved in joint regulation: 4 (coal feeder, primary air fan, secondary air fan, furnace temperature measuring system), Influence factor of process B = 2 / 4 = 0.5; Step 4: Calculate the anomaly impact coefficient and make a final judgment. Definition: Abnormal Influence Coefficient = (Number of Jointly Adjusted Deviation Processes / Total Number of Mismatch Processes) × (Average Influence Factor of All Jointly Adjusted Deviation Processes) Number of Jointly Adjusted Deviation Processes: 2, Total Number of Mismatch Processes: 2, First Part Ratio: 2 / 2 = 1.0, Average Influence Factor: (0.25 + 0.5) / 2 = 0.375, Abnormal Influence Coefficient = 1.0 × 0.375 = 0.375, Final Judgment: Abnormal Influence Coefficient (0.375) < Preset Influence Coefficient Threshold (0.7).

[0051] 4. Final Decision and Explanation Final diagnosis result: The response deviation type of the #2 boiler coal feeder intelligent agent was determined to be the matching response type.

[0052] Furthermore, determining whether the joint regulation data of agents with different response bias types meets the requirements specifically includes: Based on the joint regulation data of agents with different response bias types, the proportion of agents matching the abnormal type in different regulation processes is determined. The adjustment process that matches the anomaly type is treated as the anomaly adjustment process; Based on the constituent data of the abnormal adjustment process in the adjustment process and the proportion of agents matching the abnormal type in different adjustment processes, it is determined whether the joint adjustment data of agents with different response deviation types meets the requirements.

[0053] It is understood that, based on the constituent data of the abnormal adjustment process in the adjustment process and the proportion of agents matching the abnormality type in different adjustment processes, it is determined whether the joint adjustment data of agents with different response deviation types meets the requirements, specifically including: Based on the constituent data of the abnormal adjustment process in the adjustment process, determine the proportion of abnormal adjustment processes in the adjustment process; The joint regulation deviation factor is determined based on the average proportion of the abnormal regulation process in the regulation process and the proportion of agents matching the abnormal type in different regulation processes. Using the joint regulation deviation factor, it is determined whether the joint regulation data of agents with different response deviation types meet the requirements.

[0054] Specifically, when the joint regulation deviation factor is greater than the preset regulation deviation factor threshold, it is determined that the joint regulation data of agents with different response deviation types does not meet the requirements.

[0055] It should be noted that when the joint regulation data of agents with different response deviation types does not meet the requirements, as long as there is an updated agent, data interaction processing between the agent and other agents will be performed to realize the simulation processing of the joint regulation process in the agent, thereby ensuring the reliability of the joint regulation processing of the agents.

[0056] Specifically, multiple thermal devices within the power plant are equipped with intelligent agents. Following the initial individual diagnostic process, each agent in the system has been labeled with different response deviation types. Now, it is necessary to evaluate the overall collaborative performance of these agents during joint regulation from a system-wide perspective, determine whether their data "meets the requirements," and decide whether to initiate system-level optimization measures accordingly.

[0057] Known agents and their diagnostic results: #1 Boiler main burner agent: Matches anomaly type (assumed to be diagnosed as anomaly in the previous case), #2 Coal feeder agent: Matches response type (from the previous embodiment), #3 Forced draft fan agent: Matches response type, #4 Induced draft fan agent: Matches anomaly type, #5 Condensate pump agent: Matches response type; Anomaly Regulation Process: A regulation process is marked as an anomaly if at least one agent with a matching anomaly type participates. This implies that the "soil" for this joint regulation may be unhealthy. Joint Regulation Bias Factor: A comprehensive indicator that considers both the "frequency of anomaly occurrences" and the "density of problematic agents during anomalies."

[0058] Preset parameters: Preset adjustment deviation factor threshold: 0.4; 3. Data Preparation and Calculation Process The system analyzed the adjustment process within the most recent statistical period (e.g., 100 times).

[0059] Step 1: Statistical analysis of basic data. Total number of adjustment processes: 100. Identify abnormal adjustment processes: Examine these 100 processes. If any process involves either agent #1 or #4, which are "matching the abnormal type", it is recorded as an abnormal adjustment process.

[0060] Assuming the statistical result shows that 30 processes are labeled as anomalous regulation processes, calculate the proportion of agents matching the anomalous type: For each regulation process, calculate the proportion of agents among its participants that "match the anomalous type". For example, if #1, #2, and #3 participate in one process, the proportion of anomalous agents is 1 / 3 ≈ 0.33. If #1 and #4 participate in another process, the proportion of anomalous agents is 2 / 2 = 1.0. Calculate this proportion for each of these 100 processes, and then calculate the average of all these proportions. After calculation, this globally average proportion of agents matching the anomalous type is 0.25.

[0061] Step 2: Calculate the joint adjustment deviation factor formula: Joint regulation bias factor = the average of the proportion of abnormal regulation processes and the global average proportion of matched abnormal agents.

[0062] Calculations: Proportion of abnormal adjustment processes = 30 / 100 = 0.3, Global average proportion of matched abnormal agents = 0.25, Joint adjustment bias factor = (0.3 + 0.25) / 2 = 0.275.

[0063] Step 3: Assessment of Compliance Judgment: The joint regulation deviation factor (0.075) is greater than the preset regulation deviation factor threshold (0.1). Therefore, the system determines that "the joint regulation data of agents with different response deviation types do not meet the requirements".

[0064] Cost-effectiveness decision: Initiating deep data interaction and simulation training between agents is a resource-intensive task. Based on quantitative assessment, the system determines that this cost is not currently necessary, thus making a cost-effective decision. Operations personnel can continue to focus on monitoring and preparing updates for specific problematic agents, rather than undertaking a system-wide reconstruction. This example demonstrates how to move from individual diagnosis to system evaluation, achieving a precise balance between "individual repair" and "system optimization" in operational strategies.

[0065] Furthermore, the method for determining the data interaction strategy of the intelligent agent is as follows: The agent that adopts the target update processing strategy is taken as the target policy agent. The joint adjustment data of the target policy agent and the agent is used to determine the adjustment process of the joint adjustment processing of the target policy agent and the agent. Based on the agent data of the target update processing strategy, determine the number of agents for the target update processing strategy; The data interaction strategy of the agent is determined based on the updated data of the target policy agent, the number of agents in the target update processing strategy, and the adjustment process of the target policy agent and the agent performing joint adjustment processing.

[0066] Furthermore, based on the updated data of the target policy agent, the number of agents in the target update processing strategy, and the adjustment process of the joint adjustment processing between the target policy agent and the agent, the data interaction strategy of the agent is determined, specifically including: Based on the updated data of the target policy agent, the number of target policy agents that have been updated is determined. If the number of target policy agents that have been updated is greater than a preset threshold for the number of agents, the data interaction strategy of the agents is determined to be that all agents perform data interaction processing at the same time, thereby determining whether the updated target policy agents can meet the adjustment requirements.

[0067] Furthermore, if the number of target policy agents that need updating is not greater than a preset agent number threshold, then the number of target policy agents is obtained. If the number of target policy agents is less than the preset target agent number threshold, then the number of target policy agents is relatively small. Therefore, even if there is an update to the target policy agents, all agents will simultaneously perform data interaction processing to simulate the adjustment results between agents under different adjustment commands, thereby determining whether the updated target policy agents meet the requirements. The impact on the normal operation of the agents is also relatively low. Therefore, the data interaction strategy of the agents is determined to be that all agents perform data interaction processing simultaneously, thereby determining whether the updated target policy agents can meet the adjustment requirements.

[0068] Additionally, it should be noted that if the number of target policy agents is not less than a preset target agent number threshold, the collaborative correlation factor between the target policy agents with updates and the agents is determined based on the number of adjustment processes where updated target policy agents and the agents jointly perform adjustment processing. When the collaborative correlation factors between different agents and the target policy agents with updates are all greater than the preset correlation factor threshold, the data interaction strategy of the agents is determined to be that all agents perform data interaction processing simultaneously, thereby determining whether the updated target policy agents can meet the adjustment requirements.

[0069] Furthermore, if the collaborative association factors between different agents and the target policy agent with the updated policy are not all greater than a preset association factor threshold, the agent's associated agents in the target policy agent are determined based on the collaborative association factors. If the target policy agent with the updated policy is an associated agent, the data interaction strategy of the agent is determined to be that all agents perform data interaction processing simultaneously, thereby determining whether the updated target policy agent can meet the adjustment requirements.

[0070] Specifically, the associated agent is the target policy agent with the largest collaborative association factor among the target policy agents.

[0071] Additionally, it is understandable that if there is an updated target policy agent that does not belong to the associated agent, then it is only necessary to perform collaborative interaction processing with agents whose collaborative association factor is greater than the preset association factor threshold.

[0072] Specifically, the intelligent system of a thermal power plant contains multiple agents. Some agents have been marked by the system as "target policy agents" that require priority updates due to poor performance. Now, after these target policy agents have completed their updates, the system needs to decide how to verify their ability to collaborate with the new environment and other agents, that is, to determine the "data interaction strategy": whether to allow all agents to participate in the simulation test simultaneously, or to allow only a portion of closely related agents to participate.

[0073] System Status: Plant-wide agent set: {A1, A2, A3, A4, A5, A6}, target policy agents (marked as requiring priority updates): {A1, A2, A3} Preset parameters: Preset threshold for the number of agents (to determine whether to update in large batches): 2, Preset threshold for the number of target agents (to determine the size of the target group): 2, Preset threshold for association factor (to determine the degree of collaboration): 0.6, Preset threshold for proportion (to determine the distribution of associated agents): 0.5, Preset threshold for factor (to determine the average degree of association): 0.4.

[0074] 2. Scenario 1: Strategies for Large-Scale Updates Scenario: Assume there are currently 3 agents updating the target policy (i.e., A1, A2, and A3 have all completed the update simultaneously). Decision judgment: 3 > preset agent number threshold (2). Decision result: Due to the large number of agents updating at once, the system stability faces significant challenges. Therefore, the system adopts the most cautious strategy: all agents perform data interaction processing simultaneously. That is, A1, A2, A3, A4, A5, and A6 are all connected to the simulation environment for comprehensive joint adjustment testing to evaluate the overall system performance after large-scale updates.

[0075] 3. Scenario 2: Strategies for Small-Batch Updates and Small Target Groups Scenario: Assume that the number of target policy agents currently updating is 1 (only A1 has completed the update). And the total number of target policy agents in the entire plant is 2 (for example, only A1 and A2 are marked). At this time, the number of updates (1) ≤ the preset agent number threshold (2), and the total number of target policy agents (2) ≤ the preset target agent number threshold (2).

[0076] Decision Outcome: Since there are only a few agents to focus on regarding the "problems," and only one is updated at a time, even a comprehensive test would have a minimal impact on the system. Therefore, the system decides: all agents will process data simultaneously. A comprehensive test will be conducted to ensure everything is in order.

[0077] 4. Scenario 3: Strategies for small-batch updates and large target groups (requires calculation of collaborative correlation factors) Now we need to consider a more complex scenario that will trigger the calculation of co-correlation factors.

[0078] Scenario: There is a target policy agent that has been updated: A1 (only one has completed the update), and the total number of target policy agents is 3 ({A1, A2, A3}) > the preset target agent number threshold (2); Joint regulation history: A1 and A4 jointly regulated 45 times in 50 cycles, A1 and A5 jointly regulated 30 times in 50 cycles, and A1 and A6 jointly regulated 10 times in 50 cycles.

[0079] 4.1 Calculate the synergistic association factor Formula: Synergistic association factor = (Number of joint regulating processes) / (Total number of regulating processes within the statistical period). Calculation (using 50 cycles as an example): Synergistic association factor between A1 and A4 = 45 / 50 = 0.9, Synergistic association factor between A1 and A5 = 30 / 50 = 0.6, Synergistic association factor between A1 and A6 = 10 / 50 = 0.2 4.2 Decision-making and Results Judgment 1: Are the co-association factors of all agents with A1 greater than the threshold (0.6), A4 (0.9>0.6), A5 (0.6=0.6), and A6 (0.2<0.6)? The result is "No".

[0080] Judgment 2: Identify the associated agent of A4 (i.e., the one with the largest co-association factor).

[0081] If any of the associated agents A4, A5, or A6 belongs to A1, and its collaborative association factor with other target policy agents is less than its collaborative association factor with A1, then all agents are determined to perform data interaction processing simultaneously. Comprehensive testing is conducted to ensure completeness.

[0082] If none of the associated agents of A4, A5, and A6 belong to A1, only A1 (the updated agent), A4 (factor 0.9), and A5 (factor 0.6) will be connected to the simulation environment for testing. Agent A6 is excluded from this round of testing due to its low association degree.

[0083] This embodiment demonstrates a highly intelligent, multi-layered decision-making system. By balancing multiple dimensions such as update scale, target group size, and historical collaboration density, it dynamically selects the most suitable data interaction strategy for the verification work after each agent update. This ensures the sufficiency of system verification while minimizing unnecessary computational resource consumption and impact on the online system, achieving a balance between precision and economy.

[0084] Example 2 In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the aforementioned multi-agent-based thermal power plant operation management method when running the computer program.

[0085] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0086] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0087] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A multi-agent-based thermal power plant operation management method, characterized by, Specifically comprising: Based on the operation data of the thermal power plant, the adjustment data of different thermal equipment in the thermal power plant is determined, the updating processing strategy of the agent of the thermal equipment is determined based on the adjustment data and the joint adjustment data of the thermal equipment without agent, and when the updating processing strategy does not belong to the target updating processing strategy, the next step is entered; Based on the adaptation data of the agent of the thermal equipment and the adjustment instruction and the adaptation data of other thermal equipment in the joint adjustment process, the response deviation type of the agent of the thermal equipment is determined, and when the joint adjustment data of the agents of different response deviation types meets the requirements, the next step is entered; Based on the joint adjustment data between different agents and the agent data of the target updating processing strategy, the data interaction strategy of the agent after the updating of the agent of the target updating processing strategy is determined.

2. The multi-agent based thermal power plant operation management method according to claim 1, wherein, The thermal equipment includes a coal mill, a coal feeder, a fan, a water supply pump, a boiler, a steam turbine, and a desuperheater.

3. The multi-agent based thermal power plant operation management method according to claim 1, wherein, The adjustment data of the thermal equipment includes the number of times of adjustment processing of the thermal equipment in history.

4. The multi-agent based thermal power plant operation management method according to claim 1, wherein, The method for determining the updating processing strategy of the agent of the thermal equipment is: Based on the adjustment data, the adjustment processing process of the thermal equipment in history is determined and is taken as an adjustment processing process; According to the joint adjustment data of the thermal equipment without agent in the adjustment processing process, the thermal equipment without agent which performs joint adjustment processing in the adjustment processing process is determined and is taken as an adjustment defect thermal equipment; Based on the adjustment defect thermal equipment data in different adjustment processing processes, the updating processing strategy of the agent of the thermal equipment is determined.

5. The multi-agent based thermal power plant operation management method according to claim 4, characterized by, Based on the adjustment defect thermal equipment data in different adjustment processing processes, the updating processing strategy of the agent of the thermal equipment is determined, specifically comprising: Based on the adjustment defect thermal equipment data in different adjustment processing processes, the number of adjustment defect thermal equipment in different adjustment processing processes is determined; The adjustment processing process in which the number of adjustment defect thermal equipment does not meet the requirements is taken as a risk adjustment process; According to the risk adjustment process data, the updating processing strategy of the agent of the thermal equipment is determined.

6. The multi-agent based thermal power plant operation management method according to claim 5, wherein The risk adjustment process is an adjustment processing process in which the number of adjustment defect thermal equipment is greater than a preset defect thermal equipment number threshold.

7. The multi-agent based thermal power plant operation management method according to claim 5, wherein When the number of risk adjustment processes of the thermal equipment does not meet the requirements, the updating processing strategy of the agent of the thermal equipment is determined as a target updating processing strategy, that is, when the updating number of the adjustment processing process meets the requirements, the updating processing of the agent of the thermal equipment is performed.

8. The multi-agent based thermal power plant operation management method according to Claim 1, wherein The method for determining the data interaction strategy of the agent is: The agent adopting the target updating processing strategy is taken as a target strategy agent, and the adjustment process in which the target strategy agent and the agent perform joint adjustment processing is determined based on the joint adjustment data of the target strategy agent and the agent; According to the agent data of the target updating processing strategy, the number of agents of the target updating processing strategy is determined; Determine the data interaction strategy of the agent based on the update data of the target policy agent, the number of agents of the target update processing policy, and the adjustment process of the joint adjustment processing of the target policy agent and the agent.

9. The multi-agent based thermal power plant operation management method according to Claim 8, characterized by, Determine the data interaction strategy of the agent based on the update data of the target policy agent, the number of agents of the target update processing policy, and the adjustment process of the joint adjustment processing of the target policy agent and the agent, specifically comprising: Determine the number of target policy agents that exist for updating based on the update data of the target policy agent, wherein if the number of target policy agents that exist for updating is greater than a preset agent number threshold, it is determined that the data interaction strategy of the agent is that all agents simultaneously perform data interaction processing, thereby determining whether the updated target policy agent can meet the adjustment demand.

10. A computer system comprising: The memory and processor connected in communication, and the computer program stored on the memory and capable of running on the processor, characterized in that the processor executes the computer program to perform the method of claim 1-9.

Images