Agent-driven coal-fired boiler combustion optimization closed-loop control method and system

Abstract

The application relates to the technical field of boiler combustion optimization, and discloses an agent-driven coal-powder boiler combustion optimization closed-loop control method and system, which comprises the following steps: acquiring combustion data in a coal-powder boiler combustion process; combining the combustion data with a preset combustion task, and generating an initial combustion control strategy through a power system multi-agent hierarchical cooperation mode; optimizing the initial combustion control strategy according to the combustion data and the initial combustion control strategy, obtaining an optimized combustion strategy by predicting a boiler combustion state in a preset future period, guiding the boiler to make combustion adjustment, and performing closed-loop control on the combustion optimization process of the coal-powder boiler; through the closed-loop control on the combustion optimization process of the coal-powder boiler, different combustion tasks can be responded in time, the control strategy can be automatically adjusted according to the combustion task and the boiler combustion state, the control performance can be continuously improved, the combustion state can always meet the task demand, and the stable operation of the coal-powder boiler is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of boiler combustion optimization, in particular to an agent-driven pulverized coal boiler combustion optimization closed-loop control method and system. BACKGROUND

[0002] The traditional boiler combustion control method adopts a centralized control mode, and the overall control scheme needs to be optimized in the control mode optimization process, which has low optimization efficiency, resulting in insufficient timely optimization response to the control process. In the process of boiler combustion optimization, intelligent control and adaptive control are lacking, and it is difficult to meet the application requirements of different combustion tasks and combustion scenarios.

[0003] A boiler optimization control method based on a combustion state prediction model is disclosed in Chinese patent application No. CN115437245A, which includes a random forest machine learning algorithm and a model online updating function to build a boiler combustion state prediction model, and then obtains the optimal adjustable state parameters under the current working condition through a combustion state classification algorithm to guide the boiler to adjust the combustion, so as to achieve the combustion state self-optimization function under multi-objective optimization, and realize closed-loop optimization and control of the boiler.

[0004] The above prior art has the problems proposed in the background: the centralized control mode has low optimization efficiency, resulting in insufficient timely optimization response to the control process; and intelligent control and adaptive control are lacking, and it is difficult to meet the application requirements of different combustion tasks and combustion scenarios. To solve any of the above problems, the present application proposes an agent-driven pulverized coal boiler combustion optimization closed-loop control method and system. SUMMARY

[0005] In view of the deficiencies of the prior art, the main purpose of the present application is to provide an agent-driven pulverized coal boiler combustion optimization closed-loop control method and system, which can effectively solve the problems in the background. The specific technical scheme of the present application is as follows:

[0006] The agent-driven pulverized coal boiler combustion optimization closed-loop control method comprises:

[0007] obtaining combustion data in the combustion process of the pulverized coal boiler;

[0008] According to the combustion data and the preset combustion task, an initial combustion control strategy is generated through the multi-agent hierarchical cooperation of the power system.

[0009] According to the combustion data and the initial combustion control strategy, the initial combustion control strategy is optimized by predicting the boiler combustion state in a preset future period to obtain an optimized combustion strategy, which guides the boiler to adjust the combustion, so as to realize closed-loop control of the combustion optimization process of the pulverized coal boiler.

[0010] Specifically, according to the combustion data, an initial combustion control strategy is generated by means of multi-agent hierarchical cooperation of the power system in combination with a preset combustion task, including:

[0011] The preset combustion task is decomposed to obtain a task network;

[0012] According to the task network, a plurality of agents performing corresponding tasks are found in a preset multi-agent network to obtain an agent combination;

[0013] According to the combustion data, each agent in the agent combination formulates a corresponding control strategy to obtain the initial combustion control strategy.

[0014] Specifically, the preset combustion task is decomposed to obtain a task network, including:

[0015] The preset combustion task is decomposed to obtain a plurality of subtasks;

[0016] According to the refinement degree of the subtasks, the plurality of subtasks are processed in layers to obtain a plurality of layers of subtasks;

[0017] According to the dependency relationship between adjacent layers of subtasks, the subtasks between adjacent layers are connected to obtain a task network, wherein at least one subtask in a lower layer of subtasks is connected to a higher layer of subtasks.

[0018] Specifically, according to the task network, a plurality of agents performing corresponding tasks are found in a preset multi-agent network to obtain an agent combination, including:

[0019] According to the boiler combustion process, the multi-agent combination is initialized to obtain a preset multi-agent network;

[0020] According to the topological structure of the task network, a task adjacency matrix is calculated;

[0021] Through a preset task feature extraction model, features of each task in the task network are extracted to obtain task features;

[0022] In combination with the task adjacency matrix and the task features, a preset agent selection model is used to calculate a reward value of each agent corresponding to each task;

[0023] According to the reward value, an agent with the highest reward value corresponding to each task is selected in the preset multi-agent network from high to low according to the hierarchical relationship of the task network to obtain a plurality of agents;

[0024] According to the connection relationship of the subtasks in the task network, the plurality of agents are connected correspondingly to obtain an agent combination.

[0025] Specifically, the agent combination according to the combustion data, each agent in the agent network formulates a corresponding control strategy to obtain an initial combustion control strategy, including:

[0026] In the agent network, the highest layer agent formulates a corresponding high-level control strategy according to the corresponding task;

[0027] In the agent network hierarchy from high to low, the low-level agent between adjacent layers formulates a corresponding low-level control strategy according to the high-level control strategy of the corresponding connected high-level agent and the corresponding subtask;

[0028] Until each agent in the agent network formulates a corresponding control strategy to obtain an initial combustion control strategy.

[0029] Specifically, according to the combustion data and the initial combustion control strategy, the initial combustion control strategy is optimized by predicting the boiler combustion state in a preset future period to obtain an optimized combustion strategy, which guides the boiler to adjust the combustion to close-loop control the combustion optimization process of the pulverized coal boiler, including:

[0030] According to the combustion data and the initial combustion control strategy, the boiler combustion data in a preset future period is predicted by a preset combustion prediction model to obtain a predicted combustion state;

[0031] The predicted combustion state is compared with a preset combustion task, and the initial combustion control strategy is optimized to obtain an optimized combustion strategy;

[0032] According to the optimized combustion strategy, the combustion process of the pulverized coal boiler is controlled to realize closed-loop control of the combustion optimization process of the pulverized coal boiler.

[0033] Specifically, the agent combination according to the combustion data, each agent in the agent network formulates a corresponding control strategy to obtain an initial combustion control strategy, including:

[0034] According to the combustion data and the initial combustion control strategy, the boiler combustion data in a preset future period is predicted by a preset combustion prediction model to obtain predicted combustion data;

[0035] According to the predicted combustion data, the combustion state of the boiler is judged to obtain a predicted combustion state.

[0036] Specifically, the predicted combustion state is compared with a preset combustion task, and the initial combustion control strategy is optimized to obtain an optimized combustion strategy, including:

[0037] The predicted combustion state is compared with the preset combustion task to obtain difference combustion data in the combustion data;

[0038] The difference combustion data is used to update a combustion data sample pool to obtain an updated sample pool;

[0039] The updated sample pool is used to update a control strategy of a corresponding agent to obtain an updated control strategy;

[0040] The updated control strategy is used to optimize a control strategy of the corresponding agent in the initial combustion control strategy to obtain an optimized control strategy.

[0041] Specifically, the updated sample pool is used to update a control strategy of a corresponding agent to obtain an updated control strategy, including:

[0042] The difference combustion data added in the updated sample pool is used to find an agent corresponding to the combustion task in an agent combination to obtain an agent to be updated;

[0043] The updated sample pool is used in combination with the corresponding task to update a control strategy of the agent to be updated to obtain an updated agent control strategy;

[0044] The updated agent control strategy is used to update a control strategy of an agent connected to the agent to be updated in the agent combination to obtain an updated control strategy.

[0045] An agent-driven pulverized coal boiler combustion optimization closed-loop control system is used to implement the agent-driven pulverized coal boiler combustion optimization closed-loop control method, and includes:

[0046] A data acquisition module is configured to acquire combustion data in a pulverized coal boiler combustion process;

[0047] A control strategy formulation module is configured to generate an initial combustion control strategy by means of multi-agent hierarchical cooperation according to the combustion data in combination with a preset combustion task;

[0048] A control process optimization module is configured to optimize the initial combustion control strategy to obtain an optimized combustion strategy by predicting a boiler combustion state in a preset future period, so as to guide the boiler to adjust the combustion and to perform closed-loop control on the combustion optimization process of the pulverized coal boiler.

[0049] Compared with the prior art, the present application has the following beneficial effects:

[0050] The application optimizes the combustion control strategy by multi-agent hierarchical cooperation combined with combustion state prediction, realizes closed-loop control of the coal-fired boiler combustion optimization process, can respond to different combustion tasks in time, automatically adjusts the control strategy according to the combustion task and the boiler combustion state, continuously improves the control performance, reduces manual intervention, improves production efficiency, realizes intelligent and adaptive control, makes the combustion state always meet the task demand, improves the energy utilization rate, reduces the fuel consumption, and guarantees the stable operation of the coal-fired boiler. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 The figure is a work flow chart of the Agent-driven coal-fired boiler combustion optimization closed-loop control method in the embodiment 1 of the application;

[0052] Figure 2 The figure is a structural schematic diagram of the power system agent in the embodiment 1 of the application

[0053] Figure 3 The figure is a schematic diagram of the task network construction in the embodiment 1 of the application;

[0054] Figure 4 The figure is a schematic diagram of the agent combination construction in the embodiment 1 of the application;

[0055] Figure 5 The figure is a structural schematic diagram of the Agent-driven coal-fired boiler combustion optimization closed-loop control system in the embodiment 2 of the application. DETAILED DESCRIPTION

[0056] In order to make the above objectives, characteristics and advantages of the application more apparent, the specific implementation manners of the application will be described in detail below with the accompanying drawings.

[0057] In the following description, a lot of specific details are set forth in order to facilitate a full understanding of the application, but the application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the application, therefore the application is not limited by the specific embodiments disclosed below.

[0058] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation manner of the application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is the embodiment independent or selective and mutually exclusive with other embodiments.

[0059] Embodiment 1

[0060] The embodiment provides an Agent-driven coal-fired boiler combustion optimization closed-loop control method, as shown in the figure Figure 1As shown, the Agent-driven pulverized coal boiler combustion optimization closed-loop control method comprises:

[0061] S101, acquiring combustion data in the combustion process of the pulverized coal boiler;

[0062] S102, according to the combustion data and the preset combustion task, an initial combustion control strategy is generated by means of multi-agent hierarchical cooperation of the power system;

[0063] S103, according to the combustion data and the initial combustion control strategy, the initial combustion control strategy is optimized by predicting the boiler combustion state in a preset future period to obtain an optimized combustion strategy, which guides the combustion adjustment of the boiler to realize closed-loop control of the combustion optimization process of the pulverized coal boiler.

[0064] At present, the optimization control method of the combustion process of the pulverized coal boiler has the problems of single control mode and slow response. The embodiment combines the power system agent and the combustion state prediction to realize real-time optimization of the combustion process and closed-loop control of the combustion optimization process. The multi-agent hierarchical control mode is used to provide diversified control, decompose complex tasks, and improve the flexibility and adaptability of the system. The system can better cope with variable combustion conditions and improve the task response speed. According to the combustion state prediction, the control strategy is optimized to realize closed-loop control, so that the control is more accurate and efficient.

[0065] In the embodiment, first, the combustion data of the pulverized coal boiler is collected, and corresponding sensors are arranged at multiple key positions of the boiler, such as temperature sensors installed at different heights of the furnace to collect the furnace temperature, flue gas composition sensors installed in the flue to detect the contents of oxygen, carbon monoxide, nitrogen oxides, etc., flow sensors installed in the fuel and air pipelines to measure the fuel and air flow, and pressure and water level sensors installed in the steam drum to monitor the steam pressure and water level. The combustion data of the boiler is collected to provide a data basis for the analysis of the combustion state of the boiler.

[0066] Specifically, according to the combustion data and the preset combustion task, a plurality of power system agents are used to formulate corresponding combustion control strategies. The combustion task usually has complexity and multi-target, such as improving combustion efficiency, reducing pollutant emission, and maintaining stable output power. The complex task is decomposed into multiple sub-tasks, and each agent is responsible for a corresponding sub-task. The high-level agent considers the task and the overall combustion condition from a global perspective to formulate a macro strategy. The bottom agent executes specific control operations according to the high-level strategy and local combustion data. Through this hierarchical cooperation, the advantages of each agent can be fully utilized to enhance the adaptability of the control strategy. Different agents can be flexibly adjusted according to the sub-tasks and real-time data they are responsible for.

[0067] wherein, as Figure 2 The power system agent is an integrated intelligent control system, which mainly receives different data from the power system and specific requirements, analyzes and outputs the processing results. The power system agent includes a data processing classification layer and multiple data processing modules. Each data processing module is configured with different types of models and can process input data according to specific task requirements. The task of the data processing classification layer in this architecture is to analyze the input data of the model, input the data into the corresponding processing module according to the requirements of different tasks, and ensure that the task can be effectively processed and accurate processing results can be obtained.

[0068] Specifically, according to the combustion data and the formulated combustion control strategy, the combustion state in the future period of time is predicted. Through the prediction of the combustion state in the future period of time, potential problems and trends can be found in advance, so as to optimize the initial combustion control strategy. The optimized strategy can better adapt to the future working condition change, realize closed-loop control, continuously adjust the strategy, realize the continuous optimization of the pulverized coal boiler combustion process, improve the combustion efficiency and reduce the pollutant emission.

[0069] For example, a pulverized coal boiler is used for power peak regulation. The preset combustion task is to maintain the combustion efficiency higher than 90% while meeting the rapid change of power grid load, and the nitrogen oxide emission is lower than 50ppm. Through the sensors installed in each part of the pulverized coal boiler, real-time collection of fuel flow, air flow, combustion chamber temperature, pressure, and pollutant content in exhaust gas and other data is realized. The high-level agent formulates a macro strategy according to the collected data and the corresponding responsible task, including increasing the supply proportion of fuel and air in advance during the load rising stage and other strategies. The low-level agent obtains the implementation strategy of the specific implementation mode according to the high-level strategy and the combustion data, including adjusting the fuel supply amount and air flow and other strategies.

[0070] Further, the pre-trained LSTM model is used to predict the combustion data in the next 10 minutes according to the control strategy and the combustion data, including output power, combustion efficiency and pollutant emission. If it is predicted that the combustion efficiency will drop to 85% and the nitrogen oxide emission will rise to 60ppm when the load rises rapidly, the initial combustion control strategy is optimized and adjusted in real time according to the predicted combustion data, the mixing ratio of fuel and air and the combustion time are adjusted, the optimized combustion strategy is obtained, the optimized combustion strategy is sent to the actuator to adjust the combustion process, and the combustion data is continuously monitored, the above prediction and optimization process is repeated, and the continuous optimization and stable operation of the pulverized coal boiler combustion are realized.

[0071] The application optimizes the combustion control strategy by multi-agent hierarchical cooperation combined with combustion state prediction, realizes closed-loop control of the coal-fired boiler combustion optimization process, can respond to different combustion tasks in time, automatically adjusts the control strategy according to the combustion task and the boiler combustion state, continuously improves the control performance, reduces manual intervention, improves production efficiency, realizes intelligent and adaptive control, makes the combustion state always meet the task demand, improves the energy utilization rate, reduces the fuel consumption, and guarantees the stable operation of the coal-fired boiler.

[0072] Further, according to the combustion data combined with the preset combustion task, an initial combustion control strategy is generated by the multi-agent hierarchical cooperation of the power system, including:

[0073] S201, task decomposition is performed on the preset combustion task, and a task network is constructed;

[0074] S202, according to the task network, a plurality of agents performing corresponding tasks in a preset multi-agent network are found, and an agent combination is obtained;

[0075] S203, according to the combustion data, each agent in the agent combination formulates a corresponding control strategy, and an initial combustion control strategy is obtained.

[0076] The current combustion process control method formulates a control strategy that is not detailed enough, and the strategy formulation process takes a long time, resulting in a less timely task response; the embodiment combines task network construction and agent matching, decomposes the task, each agent is responsible for a corresponding subtask, the formulated control strategy is more detailed, and the task is reasonably distributed and the agents are efficiently cooperated, improving the overall performance of the system and the response ability to the working condition change.

[0077] In the embodiment, first, the input combustion task is decomposed into a plurality of subtasks, the complex problem is simplified, a task network is constructed according to the correlation between the subtasks, the dependency relationship, the sequence and the logical connection between the subtasks are clearly shown, the task execution is avoided, and the work efficiency is improved.

[0078] Specifically, according to the task network, a matched agent combination is found in the preset multi-agent network, each agent in the preset multi-agent network has a corresponding responsible function, corresponding to the subtasks in the task network, for example, a fuel control agent is responsible for tasks such as fuel delivery and flow regulation; a wind volume control agent focuses on air flow and pressure regulation; by matching the subtasks in the task network and the agents in the multi-agent network, an agent suitable for executing each subtask can be found, an agent combination is formed, the reasonable division of labor of the agents is realized, each agent is responsible for its good task, and the accuracy and efficiency of task execution are improved.

[0079] Specifically, each agent in the agent combination formulates a corresponding control strategy according to the sub-tasks it is responsible for and the obtained combustion data, using its internal data processing module and decision model, which can improve the pertinence of the control strategy, and each agent formulates a strategy for the sub-tasks it is responsible for, which is more in line with actual needs.

[0080] Further, the task network is constructed by task decomposition of the preset combustion task, including:

[0081] S301, decomposing the preset combustion task to obtain a plurality of sub-tasks;

[0082] S302, according to the refinement degree of the sub-tasks, the plurality of sub-tasks are processed in layers to obtain a plurality of layers of sub-tasks;

[0083] S303, according to the dependency relationship between adjacent layers of sub-tasks, the sub-tasks between adjacent layers are connected to obtain a task network, wherein at least one sub-task in the low layer sub-tasks and the high layer sub-tasks is connected.

[0084] In this embodiment, as Figure 3 The combustion task is decomposed, and the preset combustion task is decomposed into a plurality of sub-tasks according to the task function, for example, the task is decomposed into a fuel supply sub-task, which is responsible for accurately controlling the delivery amount and delivery speed of fuel; an air supply sub-task, which ensures that the appropriate proportion of air is mixed with fuel; a burner control sub-task, which adjusts the angle, flame shape and other parameters of the burner to optimize the combustion effect; a pollution emission control sub-task, which reduces the generation of nitrogen oxides and other pollutants by adjusting the combustion conditions; through task decomposition, the complex combustion task is refined into simple sub-tasks, reducing the processing difficulty, each sub-task has clear work content, facilitating the allocation of resources and responsibilities, and improving work efficiency.

[0085] Specifically, the plurality of sub-tasks after decomposition are processed in layers, and the refinement degrees of different sub-tasks are different, for example, the sub-task of "improving combustion efficiency" belongs to a general task and has a greater impact on the overall task, while the sub-task of "adjusting the speed of the fuel pump" is more specific and is a specific operation to implement the combustion strategy. According to the refinement degree of the sub-tasks, the sub-tasks are divided into high layers, which involve overall target planning and strategy formulation; the sub-tasks are divided into low layers, which involve specific device operation and parameter adjustment; the sub-tasks between the two are divided into middle layers; a clear task hierarchy is formed, which facilitates task allocation and management and improves management efficiency. Different levels of sub-tasks can be independently adjusted and optimized to adapt to different combustion conditions and task requirements.

[0086] Specifically, each subtask in the combustion task does not exist in isolation, and there is a corresponding dependency relationship between tasks. The logical relationship between adjacent layer subtasks is analyzed, it is determined which subtasks have a dependency relationship, the subtasks between adjacent layers are connected using directed edges according to the corresponding dependency relationship, a task network is constructed, and the entire combustion control process forms an organic whole; the order and coordination of task execution are ensured through subtask connection, the goal of a high-level subtask is achieved through specific operations of a low-level subtask, and the execution of a low-level subtask is guided and constrained by a high-level subtask.

[0087] Further, according to the task network, a plurality of agents performing corresponding tasks are found in a preset multi-agent network, and an agent combination is obtained, including:

[0088] S401, initializing the multi-agent combination according to the boiler combustion process, and obtaining a preset multi-agent network;

[0089] S402, calculating a task adjacency matrix according to the topological structure of the task network;

[0090] S403, extracting features of each task in the task network through a preset task feature extraction model, and obtaining task features;

[0091] S404, combining the task adjacency matrix and the task features, and calculating a reward value of each agent corresponding to each task through a preset agent selection model;

[0092] S405, selecting an agent corresponding to each task with the highest reward value in the preset multi-agent network according to the reward value and the task network hierarchical relationship from high to low, and obtaining a plurality of agents;

[0093] S406, connecting the plurality of agents according to the connection relationship of the subtasks in the task network, and obtaining an agent combination.

[0094] In this embodiment, as Figure 4 , the performance of the optimal agent for each subtask in the task network is matched, the selected agents are combined according to the corresponding relationship, an agent combination is obtained, each agent in the agent combination cooperates according to the dependency relationship of the task, the performance of the optimal agent for each task is directly selected, the training process is reduced, the efficiency of agent selection and calculation is improved, resource waste and task conflicts are avoided, and the running efficiency of the system is improved.

[0095] Specifically, based on the combustion process of a pulverized coal boiler, which involves multiple complex stages and functions including fuel supply, air delivery, combustion reaction, heat transfer, steam generation, and pollutant emission control, each stage requires specific control and decision-making. Through analysis of these stages and functions, intelligent agents are initialized and pre-trained. A pre-defined multi-agent network is constructed based on these initialized agents, resulting in an initial multi-agent combination. Each agent possesses corresponding implementation functions and efficiencies, achieving distributed intelligent control and improving the system's flexibility and adaptability. Based on the physical process of boiler combustion and considering the types and functions of the agents, several corresponding candidate agents are set for each sub-task, resulting in an agent set.

[0096] Based on the topology of the task network, an adjacency matrix is ​​constructed to quantify the dependencies between tasks in the task network. A parameter vector and a corresponding adjacency matrix are initialized for each agent. Based on the static and dynamic features of the task scenario, features are extracted using a pre-defined task feature extraction model. This model can be a random forest model, a convolutional neural network model, or a recurrent neural network model. In this embodiment, a random forest model is used. A pre-trained task feature extraction model is obtained by training the random forest model with a large amount of historical data. Task features are calculated using this pre-trained model, and a task feature vector is constructed. Combining the task adjacency matrix and task features, a pre-defined agent selection model is used to calculate the reward value for each agent for each task. Specifically, the agent selection model is a context-based multi-armed slot machine algorithm model. This model can dynamically select the agent with the best performance based on real-time task scenario features without pre-training. The reward value for each agent is calculated using the agent selection model, and the calculation formula is as follows:

[0097] ;

[0098] In the formula, Let i be the reward value for agent i. For task feature vectors, Let i be the parameter vector of agent i. Adjacency matrix The element in the middle represents the connection relationship between node k and node l. To control the intensity of exploration (which can be set according to the required calculation accuracy). Task feature vector The k-th element in Task feature vector The l-th element.

[0099] According to the calculated reward value of each agent, the agent with the highest reward value corresponding to the task is selected from the highest layer of the task network structure in sequence, and the agent with a high reward value corresponds to a strong performance in processing the task. By selecting the agent with a high reward value, the agent selection process can be optimized, and the task pertinence can be improved. The agent corresponding to each task is selected from the highest layer to the lower layer in sequence, and a plurality of agents are obtained. According to the connection relationship of the subtasks in the task network, the plurality of agents are connected correspondingly, and the agent combination is obtained, so that the agents can orderly transmit information and execute tasks, and efficient distributed control is realized. After the high-level agent strategy is formulated, the message is transmitted to the corresponding connected low-level agent, and the low-level agent executes the corresponding task.

[0100] Further, each agent in the agent combination formulates a corresponding control strategy according to the combustion data to obtain an initial combustion control strategy, including:

[0101] S501, in the agent network, the highest layer agent formulates a corresponding high-level control strategy according to a corresponding task;

[0102] S502, in the agent network hierarchical structure, from high to low, the low-level agent between adjacent layers formulates a corresponding low-level control strategy according to the high-level control strategy of the corresponding connected high-level agent and the corresponding subtask;

[0103] S503, until each agent in the agent network formulates a corresponding control strategy, and an initial combustion control strategy is obtained.

[0104] The current centralized decision-making mode has slow task response speed and poor adaptability. The agent combination with hierarchical structure is used to control the boiler combustion process in real time in the embodiment, the control rate is improved, and the control strategy of each agent can be quickly adjusted according to different working conditions and task requirements, so that the flexibility and adaptability are high.

[0105] In the embodiment, the agent combination formulates an overall control strategy from the highest layer agent according to the hierarchical structure, and the corresponding low-level agent formulates a corresponding subtask control strategy according to the control strategy of the corresponding connected high-level agent, until each agent obtains a corresponding control strategy.

[0106] Specifically, the highest-level agent formulates the control strategy. The highest-level agent is in a macro decision-making position in the agent network and can control the entire boiler combustion process from a global perspective. The tasks it is responsible for are usually more macro goals in the preset combustion task, such as meeting the power grid load demand, improving overall combustion efficiency, and achieving specific pollutant emission indicators. The highest-level agent formulates a high-level control strategy according to the tasks it is responsible for and the combustion data it obtains, and provides direction and framework for the decision-making of lower-level agents through internal data processing modules.

[0107] Specifically, the bottom-level agent formulates further decisions and strategies based on the macro strategy formulated by the high-level agent, its own subtasks, and real-time combustion data. The low-level agent receives the high-level control strategy sent by the corresponding connected high-level agent and interprets it to determine the constraints and goals it needs to follow. At the same time, it analyzes the subtasks in detail to determine the specific requirements and goals of the subtasks. For example, the air volume control agent is responsible for adjusting the amount of air entering the furnace. It needs to analyze the current combustion state, fuel supply, and other factors to determine the appropriate air adjustment target. Based on the high-level control strategy, subtask requirements, and real-time combustion data, it formulates a corresponding control strategy. The control strategy is formulated layer by layer downward, and each agent ultimately formulates a corresponding control strategy based on its own task and the upper-level strategy to obtain an initial combustion control strategy. Through hierarchical transmission and integration, the consistency and coordination between the control strategies of various agents are ensured, and strategy conflicts and contradictions are avoided.

[0108] Further, based on the combustion data and the initial combustion control strategy, the initial combustion control strategy is optimized to obtain an optimized combustion strategy by predicting the boiler combustion state in a preset future period, which guides the boiler to adjust the combustion to control the coal-fired boiler combustion optimization process in a closed loop, including:

[0109] S601, predicting the boiler combustion data in a preset future period based on the combustion data and the initial combustion control strategy through a preset combustion prediction model to analyze and obtain a predicted combustion state;

[0110] S602, comparing the predicted combustion state with a preset combustion task to optimize the initial combustion control strategy to obtain an optimized combustion strategy;

[0111] S603, controlling the combustion process of the coal-fired boiler according to the optimized combustion strategy to realize closed-loop control of the coal-fired boiler combustion optimization process.

[0112] Further, according to the initial combustion control strategy, the future combustion state of the boiler is predicted, and the control strategy is optimized in advance, which overcomes the limitation of the traditional control method that can only adjust according to the current state, improves the foresight and accuracy of the control, realizes the closed-loop control of the combustion optimization process, effectively deals with various disturbances and working condition changes, maintains the stability of the boiler operation, and reduces the probability of failure.

[0113] In the embodiment, the preset combustion prediction model is an LSTM model, the LSTM model is trained according to a large amount of historical combustion data and corresponding combustion state data, and a pre-trained LSTM model is obtained; the preprocessed current combustion data and the initial combustion control strategy are input into the trained combustion prediction model, the model calculates and reasons according to the input data, and outputs the predicted value of the boiler combustion data in the preset future period, the predicted value including key parameters such as furnace temperature, flue gas temperature, flue gas composition, and steam pressure; the combustion state of the boiler is judged according to the predicted combustion data, the future combustion state of the boiler is determined, and the combustion state usually includes different categories such as good, normal, and abnormal (such as insufficient combustion, overheating, and overcooling).

[0114] Specifically, the predicted combustion state reflects the future operation of the boiler under the initial combustion control strategy, the predicted combustion state is compared with the preset combustion task, it is judged whether the initial combustion control strategy can meet the requirements of the preset combustion task, if there is a deviation, the initial combustion control strategy needs to be adjusted and optimized, so that the future combustion state is closer to the preset target; by timely adjusting the control strategy, the combustion process of the boiler is closer to the preset combustion task target, the combustion efficiency is improved, and the pollutant emission is reduced; according to different prediction results and actual conditions, the control strategy is flexibly adjusted, the adaptability and stability of the system are enhanced.

[0115] Specifically, according to the optimized combustion control strategy, the combustion process of the boiler is controlled and adjusted in real time, the optimization and adjustment process is repeatedly adjusted, a closed-loop control cycle is formed, the combustion process of the boiler is always optimized towards the preset combustion task target, the deviation in the combustion process can be corrected in time, the stability of the boiler operation is maintained, and the parameter fluctuation is reduced.

[0116] Further, the boiler combustion data in the preset future period is predicted according to the combustion data and the initial combustion control strategy through the preset combustion prediction model, and a predicted combustion state is analyzed and obtained, including:

[0117] S701, according to the combustion data and the initial combustion control strategy, the boiler combustion data in the preset future period is predicted through the preset combustion prediction model, and predicted combustion data is obtained;

[0118] S702, judging a combustion state of the boiler according to the predicted combustion data, to obtain a predicted combustion state.

[0119] In this embodiment, the boiler combustion process has continuity and regularity, according to the current combustion data and the initial combustion control strategy, the change of each parameter of the boiler in the future period can be predicted, according to the pre-trained LSTM model, the predicted value of the boiler combustion data in the future period (such as 5 minutes, 10 minutes, etc.) is predicted, and the execution effect of the initial combustion control strategy is judged and adjusted by predicting the future combustion data in advance.

[0120] Specifically, according to the predicted combustion data, the combustion state of the boiler is judged, and the future combustion state of the boiler is determined. The combustion state usually includes different categories such as good, normal, abnormal (such as insufficient combustion, overheating, overcooling, etc.). By judging the future combustion state of the boiler, the future combustion problem can be found in time, which is convenient for adjusting and optimizing the combustion process in advance and avoiding problem deterioration.

[0121] Further, the predicted combustion state is compared with the preset combustion task, and the initial combustion control strategy is optimized to obtain an optimized combustion strategy, comprising:

[0122] S801, comparing the predicted combustion state with the preset combustion task to obtain difference combustion data in the combustion data;

[0123] S802, updating the combustion data sample pool according to the difference combustion data to obtain an updated sample pool;

[0124] S803, updating the control strategy of the corresponding agent according to the updated sample pool to obtain an updated control strategy;

[0125] S804, optimizing the control strategy of the corresponding agent in the initial combustion control strategy according to the updated control strategy to obtain an optimized control strategy.

[0126] In this embodiment, the difference data obtained by comparing the predicted state with the preset task is used to update the sample pool, and the agent is optimized to obtain the control strategy, which realizes dynamic optimization based on real-time data. According to the difference data, the control strategy of the corresponding agent is adjusted, which improves the adaptability and synergy of the whole control strategy system, instead of the traditional single adjustment to the whole, which is more accurate and efficient.

[0127] Specifically, by comparing the preset combustion task with the predicted combustion state, the gap between the actual prediction and the expected target is obtained, which is reflected in the form of difference combustion data, reflecting which aspects of the boiler combustion process under the current control strategy fail to meet the preset task requirements, for example, if the preset combustion efficiency target is 90% and the predicted combustion efficiency is 85%, the difference of this indicator is -5%, and arranging the differences of all indicators into difference combustion data can quickly and accurately find out the deviation of the boiler combustion process from the preset task.

[0128] Specifically, the difference combustion data is put into the sample pool and integrated with the original data in the sample pool. The integrated data is cleaned to remove existing noise, outliers and duplicate data. At the same time, the data is preprocessed such as normalization and standardization to improve the quality and availability of the data, and an updated sample pool is obtained. According to the updated sample pool, the corresponding intelligent agent related to the difference data is retrained or its control parameters are adjusted, so that it can learn new knowledge in the updated sample pool, and an updated control strategy more suitable for the current combustion working condition is obtained. Replacing the control strategy of the corresponding intelligent agent in the initial combustion control strategy with the updated control strategy can ensure that the entire control strategy system can better cope with the current combustion condition, thereby realizing the optimization of the initial combustion control strategy.

[0129] Further, according to the updated sample pool, the control strategy of the corresponding intelligent agent is updated to obtain an updated control strategy, comprising:

[0130] S901, according to the newly added difference combustion data in the updated sample pool, find the intelligent agent corresponding to the combustion task in the intelligent agent combination, and obtain the intelligent agent to be updated;

[0131] S902, update the control strategy of the intelligent agent to be updated by using the updated sample pool in combination with the corresponding task, and obtain an updated intelligent agent control strategy;

[0132] S903, according to the updated intelligent agent control strategy, the intelligent agent connected to the intelligent agent to be updated in the intelligent agent combination is updated in the corresponding control strategy, and an updated control strategy is obtained.

[0133] In this embodiment, each agent in the agent combination is responsible for a specific combustion task, and the newly added differential combustion data in the sample pool reflects the deviation of the combustion state. Different deviations correspond to problems in different combustion task links. By analyzing the combustion task corresponding to the newly added differential combustion data, the agent that needs to update the control strategy, i.e., the agent to be updated, can be accurately located. In this way, unnecessary updates to the control strategies of all agents can be avoided, and the pertinence and efficiency of the updates can be improved. For example, when the newly added differential data shows that the fuel is not fully combusted, the corresponding combustion task is fuel supply and air ratio, etc. The control strategy of the agent is updated in a targeted manner, making the updating process more efficient and enabling the system to respond more quickly to changes in combustion conditions.

[0134] Specifically, the agent to be updated is updated according to the newly added combustion data and differential information in the update sample pool, so that the agent to be updated can learn new data patterns and rules, thereby adjusting its control strategy to obtain an updated agent control strategy. By continuously updating using the update sample pool, the control strategy of the agent is continuously optimized, and the overall performance of the combustion process is improved.

[0135] Specifically, the change of the control strategy of the agent to be updated will affect the working environment and task requirements of the agents connected thereto. According to the updated agent control strategy, the control strategies of the connected agents are adjusted accordingly to ensure the collaborative working effect of the entire agent combination, so that the entire agent control system can operate more stably, and the reliability and safety of the combustion process are improved.

[0136] Embodiment 2

[0137] In this embodiment, as Figure 5 , an agent-driven pulverized coal boiler combustion optimization closed-loop control system is provided for implementing the agent-driven pulverized coal boiler combustion optimization closed-loop control method, comprising:

[0138] a data acquisition module for acquiring combustion data during the combustion process of the pulverized coal boiler;

[0139] a control strategy formulation module for generating an initial combustion control strategy by means of multi-agent hierarchical cooperation according to the combustion data in combination with a preset combustion task;

[0140] a control process optimization module for optimizing the initial combustion control strategy by predicting the boiler combustion state in a preset future period to obtain an optimized combustion strategy, thereby guiding the boiler to adjust the combustion and performing closed-loop control on the combustion optimization process of the pulverized coal boiler.

[0141] In this embodiment, the data acquisition module includes a sensor group, a data acquisition unit, a data transmission unit and a data storage unit, which can collect various key data in the combustion process of the pulverized coal boiler in real time and accurately, and transmit and store them, through continuous acquisition of combustion data, the system can timely understand the running state of the boiler, find potential problems and change trend, so as to realize effective monitoring and optimization of the combustion process; the control strategy formulation module includes a task decomposition unit, a task network construction unit, an intelligent agent management unit, an intelligent agent matching unit and a strategy generation unit, according to the preset combustion task of the pulverized coal boiler and the real-time combustion data, through the multi-agent hierarchical cooperation mode, the initial combustion control strategy is generated, the module decomposes the complex combustion task into multiple sub-tasks, and matches the corresponding intelligent agent, fully gives play to the advantages of each intelligent agent, realizes fine control of the combustion process, and the generated initial combustion control strategy provides a guidance direction for the combustion process, which is an important link to realize combustion optimization.

[0142] Specifically, the control process optimization module includes a combustion prediction unit, a comparative analysis unit, a sample pool updating unit, an intelligent agent strategy updating unit and a control execution unit, which optimizes the initial combustion control strategy and realizes closed-loop control of the combustion process of the pulverized coal boiler, the module predicts the future combustion state, compares the preset task, updates the sample pool and the intelligent agent control strategy, and continuously adjusts the combustion control strategy, so that the combustion process of the boiler can better meet the requirements of the preset task, through closed-loop control, real-time response to the change of the combustion condition, continuous optimization of the combustion process, improvement of the combustion efficiency, reduction of the pollutant emission, and ensuring the safe, stable and efficient operation of the pulverized coal boiler.

[0143] The basic principles and main features of the present application and the advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, the above embodiments and the description in the specification are only to illustrate the principles of the present application, and various changes and improvements can be made to the present application without departing from the spirit and scope of the present application, and these changes and improvements all fall within the scope of the present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. An agent-driven closed-loop control method for combustion optimization in pulverized coal boilers, characterized in that, include: To acquire combustion data during the combustion process of a pulverized coal boiler; Based on the combustion data and the preset combustion task, an initial combustion control strategy is generated through hierarchical collaboration among multiple agents in the power system. Based on the combustion data and the initial combustion control strategy, the initial combustion control strategy is optimized by predicting the boiler combustion state in a preset future time period to obtain an optimized combustion strategy, which guides the boiler to make combustion adjustments, so as to achieve closed-loop control of the combustion optimization process of the pulverized coal boiler. Based on the combustion data and a preset combustion task, an initial combustion control strategy is generated through a multi-agent hierarchical collaboration approach within the power system, including: The preset combustion task is broken down into multiple sub-tasks; Based on the level of detail of the subtasks, multiple subtasks are processed in layers to obtain multi-layered subtasks; Based on the dependencies between adjacent subtasks, the subtasks between adjacent layers are connected to obtain a task network, wherein the lower-level subtasks in the adjacent layers are connected to at least one subtask in the higher-level subtasks. Based on the boiler combustion process, the multi-agent combination is initialized to obtain a preset multi-agent network; Calculate the task adjacency matrix based on the topology of the task network; By using a preset task feature extraction model, feature extraction is performed on each task in the task network to obtain task features; Combining the task adjacency matrix and task features, the reward value for each agent for each task is calculated using a preset agent selection model. Based on the reward value, the agent with the highest reward value for each task is selected from the preset multi-agent network in descending order of task network hierarchy, thus obtaining multiple agents; Based on the connection relationships of subtasks in the task network, the multiple agents are connected accordingly to obtain an agent combination; Based on the combustion data, each agent in the agent group formulates a corresponding control strategy to obtain an initial combustion control strategy.

2. The agent-driven closed-loop control method for pulverized coal boiler combustion optimization according to claim 1, characterized in that, Based on the combustion data, each agent in the agent group formulates a corresponding control strategy to obtain an initial combustion control strategy, including: In an agent network, the highest-level agent formulates corresponding high-level control strategies based on the specific tasks. In the hierarchical structure of the agent network, from high to low, the lower-level agents between adjacent layers formulate control strategies based on the high-level control strategies of the corresponding connected higher-level agents and in combination with the corresponding sub-tasks, thus obtaining the corresponding lower-level control strategies. The initial combustion control strategy is obtained when each agent in the agent network formulates its own control strategy.

3. The agent-driven closed-loop control method for pulverized coal boiler combustion optimization according to claim 1, characterized in that, Based on the combustion data and the initial combustion control strategy, the initial combustion control strategy is optimized by predicting the boiler combustion state in a preset future time period to obtain an optimized combustion strategy. This optimizes the boiler combustion adjustment process, enabling closed-loop control of the pulverized coal boiler's combustion optimization process, including: Based on combustion data and initial combustion control strategy, the boiler combustion data for a preset future period is predicted using a preset combustion prediction model, and the predicted combustion state is obtained through analysis. The predicted combustion state is compared with the preset combustion task, and the initial combustion control strategy is optimized to obtain an optimized combustion strategy. Based on the optimized combustion strategy, the combustion process of the pulverized coal boiler is controlled to achieve closed-loop control of the pulverized coal boiler combustion optimization process.

4. The agent-driven closed-loop control method for pulverized coal boiler combustion optimization according to claim 3, characterized in that, The process involves predicting boiler combustion data for a preset future time period based on combustion data and an initial combustion control strategy, using a preset combustion prediction model to analyze and obtain the predicted combustion state, including: Based on combustion data and initial combustion control strategy, the boiler combustion data for a preset future period is predicted using a preset combustion prediction model to obtain predicted combustion data. Based on the predicted combustion data, the combustion state of the boiler is determined to obtain the predicted combustion state.

5. The agent-driven closed-loop control method for pulverized coal boiler combustion optimization according to claim 3, characterized in that, The step of comparing the predicted combustion state with the preset combustion task and optimizing the initial combustion control strategy to obtain an optimized combustion strategy includes: The predicted combustion state is compared with the preset combustion task to obtain the difference combustion data in the combustion data; The combustion data sample pool is updated based on the differential combustion data to obtain an updated sample pool; Based on the updated sample pool, the control strategy of the corresponding agent is updated to obtain the updated control strategy; Based on the updated control strategy, the control strategy of the corresponding agent in the initial combustion control strategy is optimized to obtain the optimized control strategy.

6. The agent-driven closed-loop control method for pulverized coal boiler combustion optimization according to claim 5, characterized in that, Based on the updated sample pool, the control policy of the corresponding agent is updated to obtain the updated control policy, including: Based on the newly added differential combustion data in the updated sample pool, find the agent corresponding to the combustion task in the agent combination to obtain the agent to be updated; By using the updated sample pool in conjunction with the corresponding task, the control strategy of the agent to be updated is updated to obtain the updated agent control strategy. According to the updated agent control strategy, the control strategies of the agents connected to the agent to be updated in the agent group are updated accordingly to obtain the updated control strategy.

7. An agent-driven closed-loop control system for pulverized coal boiler combustion optimization, characterized in that, A method for implementing the agent-driven closed-loop control method for pulverized coal boiler combustion optimization as described in any one of claims 1 to 6 includes: The data acquisition module acquires combustion data during the combustion process of the pulverized coal boiler. The control strategy formulation module generates an initial combustion control strategy based on the combustion data and a preset combustion task through multi-agent hierarchical collaboration. The control process optimization module optimizes the initial combustion control strategy based on the combustion data and the initial combustion control strategy by predicting the boiler combustion state in a preset future time period, thereby obtaining an optimized combustion strategy to guide the boiler to make combustion adjustments, so as to perform closed-loop control of the combustion optimization process of the pulverized coal boiler.

Images