A full-process autonomous decision-making method for a multi-process system of a coal-fired power plant

Abstract

The present application relates to the field of intelligent control technology of coal-fired power plant, disclose a kind of coal-fired power plant multi-process system whole autonomous decision-making method, wherein the method comprises: based on mechanism model optimization and big data mining two kinds of driving mode fusion constructs autonomous decision-making algorithm;To burn, water supply, wind smoke, environmental protection and auxiliary machine five kinds of system generates collaborative decision-making instruction and eliminates control conflict;Through multi-objective optimization algorithm, the whole autonomous cruise is carried out to key parameter setting value;After safety check, the instruction and setting value are issued to DCS control system to complete whole-process autonomous execution.The present application solves the technical problems that control conflict exists between multi-process systems of coal-fired power plant, cross-system parameter consistency is difficult to guarantee, and whole-process autonomous decision cannot be realized under complex working conditions.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of intelligent control technology for coal-fired power plants, and in particular to a method, system, electronic device, and storage medium for autonomous decision-making throughout the multi-process system of a coal-fired power plant. Background Technology

[0002] As the power system continues to increase its requirements for the flexibility of coal-fired power units, the stable operation of coal-fired power plants under complex conditions such as wide load peak shaving, frequent coal quality switching and multi-pollutant coordinated emission reduction faces increasingly severe challenges. The coupling between various process systems has also deepened significantly with the increase in unit capacity and the diversification of control objectives.

[0003] The operational decisions of existing coal-fired power plants usually rely on operators to manually adjust the parameters of each process system based on experience, or adopt a local automatic control scheme for a single process system. Some schemes introduce expert systems based on fixed rules or single data-driven models to assist in parameter tuning under specific operating conditions.

[0004] However, the existing solutions lack a unified conflict resolution logic when handling cross-system coordination of multi-process systems, leading to control conflicts between combustion ratio, flue gas ratio, and environmental protection injection volume. Fixed-rule expert systems struggle to adapt to coal quality fluctuations and atypical operating conditions, while single data-driven models exhibit discrete solution bias in operating condition ranges with insufficient historical data coverage, failing to guarantee continuous optimal parameter settings throughout the entire process. Furthermore, none of the existing solutions achieve autonomous decision-making without human intervention throughout the entire process from cold start-up to shutdown, and the trade-offs between safety, economy, and environmental protection objectives lack the ability to adaptively adjust with different operating stages. The inventors of this disclosure note that current multi-process system control schemes do not consider the complementary characteristics of mechanistic models and historical data in different operating condition ranges, resulting in an inability to effectively compensate for the discrete solution problem of data-driven methods in data-sparse regions, thus causing a decline in the quality of cross-system collaborative decision-making. Summary of the Invention

[0005] The purpose of this invention is to provide a method for autonomous decision-making throughout the entire process of a multi-process system in a coal-fired power plant, which solves the technical problems of control conflicts between multi-process systems in a coal-fired power plant, difficulty in ensuring cross-system parameter consistency, and inability to achieve autonomous decision-making throughout the entire process under complex operating conditions.

[0006] This disclosure provides at least one embodiment of a fully autonomous decision-making method for a multi-process system in a coal-fired power plant. The method includes: generating decision rules and parameter settings for each process system of the coal-fired power plant using two driving methods: mechanistic model optimization and big data mining; and fusing the two types of outputs into a unified autonomous decision-making algorithm through a weighted synthesis strategy, wherein the fusion weight coefficient is calculated in real-time based on the Euclidean distance between the current operating condition feature vector and the center point of historical samples; and generating separate control command sets for the combustion system, water supply system, flue gas system, environmental protection system, and auxiliary equipment system according to the unified autonomous decision-making algorithm. The system performs cross-system consistency verification on five types of control command sets through conflict detection logic, and outputs collaborative decision commands after eliminating control conflicts. Based on a multi-objective optimization algorithm, it autonomously determines the set values ​​of key parameters according to the current load, coal quality and operating conditions, and adjusts the set values ​​of key parameters in real time during unit operation. The collaborative decision commands and the set values ​​of key parameters are then sent to the distributed control system after security verification, covering the entire process of unit cold start-up, variable load peak shaving, steady-state operation and shutdown. The collaborative decision commands are generated based on the set values ​​of key parameters, and the two satisfy parameter consistency constraints.

[0007] For example, in the fully autonomous decision-making method for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure, the fusion of two types of outputs through a weighted synthesis strategy includes: assuming the Euclidean distance between the current operating condition feature vector and the center point of the nearest historical sample is... Weights output by big data mining Weights of the output of the mechanistic model optimization satisfy , ,in This refers to the distance attenuation coefficient; the final parameter setpoint of the fused output. Depend on Confirmed, among which Assign values ​​to the decision parameters output by big data mining. The decision parameters output by the mechanistic model optimization are set. Through the above adaptive weight calculation method based on Euclidean distance, the greater the distance between the current working condition and the center point of historical samples, the higher the weight of the mechanistic model optimization, thereby ensuring the reliability of decision-making in working condition intervals where historical data coverage is insufficient.

[0008] For example, in the fully autonomous decision-making method for multi-process systems of coal-fired power plants provided in at least one embodiment of this disclosure, the big data mining extracts historical operating data that meets the constraints of safety, economy and environmental protection from the historical operating database. The historical data is partitioned into operating condition zones using a clustering algorithm with operating condition feature vectors as indexes. The input of the clustering algorithm is the operating condition feature vector of the historical operating samples, and the output is the center point of each operating condition zone and the sample set to which it belongs. The mechanism model optimization is based on the thermodynamic principles of the unit, the combustion mechanism and the environmental reaction mechanism. Optimization models are established for the combustion system, feedwater system, flue gas system, environmental protection system and auxiliary system, respectively. Each optimization model takes the current operating condition parameters as input and the control parameter setpoints of each process system as output.

[0009] For example, in the fully autonomous decision-making method for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure, the conflict detection logic performs the following checks on the five types of control command sets: checks whether the oxygen balance constraint between the combustion ratio command of the combustion system and the air volume command of the flue gas system is satisfied; checks whether the flow supply capacity constraint between the feedwater flow command of the feedwater system and the feedwater pump start / stop command of the auxiliary equipment system is satisfied; checks whether the ammonia-nitrogen molar ratio constraint between the ammonia injection command of the environmental protection system and the flue gas flow command of the flue gas system is satisfied; when any constraint is not satisfied, the commands of the relevant systems are iteratively corrected until all constraints are satisfied, and then the collaborative decision command is output.

[0010] For example, in the fully autonomous decision-making method for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure, the multi-objective optimization algorithm uses safety objectives, economic objectives, and environmental objectives to form a three-dimensional objective function vector. The decision variable vector consists of main steam pressure, main steam temperature, reheat steam temperature, furnace oxygen content, denitrification setpoint, and desulfurization setpoint. The Pareto optimal frontier is solved within the safe operating range of each parameter, and the target weight vector is used as the basis for the solution. Select the parameter set with the highest weighted composite score from the Pareto front. As a fixed value output for the aforementioned key parameter; wherein... For security target components, As an economic objective component, As an environmental protection goal, , , These are the corresponding weights, satisfying... .

[0011] For example, in the fully autonomous decision-making method for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure, the target weight vector Adaptive adjustment based on the current operating phase of the unit: during the variable load peak shaving phase. Improvement; during the steady-state operation phase, Improvement; when environmental emission restriction warnings are triggered, promote.

[0012] For example, in the fully autonomous decision-making method for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure, the safety verification includes: comparing each collaborative decision-making instruction and key parameter setting with a preset safety boundary, truncating instructions or settings that exceed the safety boundary to the nearest safety boundary value and recording the truncation event; for auxiliary machine instructions involving equipment start-up and shutdown, verifying whether the current equipment status meets the preconditions for start-up and shutdown, and temporarily suspending the issuance of instructions that do not meet the preconditions and triggering waiting logic; when an abnormal operating condition detection signal is triggered, automatically switching to a preset safety decision-making mode, generating a minimum set of intervention instructions to ensure the safe operation of the unit and issuing them with priority.

[0013] At least one embodiment of this disclosure also provides a fully autonomous decision-making system for a multi-process system in a coal-fired power plant. The fully autonomous decision-making system for a multi-process system in a coal-fired power plant includes a decision algorithm construction unit, a collaborative decision-making unit, a setpoint cruise unit, and an autonomous execution unit. The decision algorithm construction unit is configured to generate decision rules and parameter setpoints for each process system of the coal-fired power plant based on two driving methods: mechanistic model optimization and big data mining. The two types of outputs are then fused into a unified autonomous decision-making algorithm using a weighted synthesis strategy. The fusion weight coefficient is calculated in real-time based on the Euclidean distance between the current operating condition feature vector and the center point of historical samples. The collaborative decision-making unit is configured to, according to the unified autonomous decision-making algorithm, separately process the combustion system, water supply system, and air supply system. The smoke control system, environmental protection system, and auxiliary equipment system generate their respective sets of control commands. A conflict detection logic is used to perform cross-system consistency verification on the five sets of control commands, eliminating control conflicts and outputting collaborative decision commands. The setpoint cruise unit is configured to autonomously determine key parameter setpoints based on a multi-objective optimization algorithm, according to the current load, coal quality, and operating conditions, and adjust these key parameter setpoints in real time during unit operation. The autonomous execution unit is configured to send the collaborative decision commands and the key parameter setpoints to the distributed control system after security verification, covering the entire process of unit cold start-up, variable load peak shaving, steady-state operation, and shutdown. The collaborative decision commands are generated based on the key parameter setpoints, and both satisfy parameter consistency constraints.

[0014] At least one embodiment of this disclosure also provides an electronic device, including: at least one memory storing computer-executable instructions non-transitory; and at least one processor configured to run the computer-executable instructions, wherein the computer-executable instructions, when run by the processor, implement the fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to any embodiment of this disclosure.

[0015] At least one embodiment of this disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer-executable instructions, which, when executed by at least one processor, implement the fully autonomous decision-making method for a multi-process system of a coal-fired power plant as described in any embodiment of this disclosure.

[0016] The beneficial effects of this invention are as follows: This invention solves the technical problem of decision parameter deviation under sparse data or atypical working conditions by integrating mechanism model optimization with big data mining with distance adaptive weights. It achieves the technical effect of outputting continuous and self-consistent parameter values ​​in the entire working condition range. By introducing cross-system conflict detection logic to perform consistency verification on the instruction sets of five types of process systems, the technical problem of constraint conflicts such as oxygen balance, flow supply capacity and ammonia-nitrogen molar ratio among control instructions of multiple process systems is solved, and the technical effect of collaborative decision-making instructions can be issued only after all cross-system constraints are satisfied. By combining a multi-objective optimization algorithm with an adaptive weight vector for the operation phase to solve the key parameter setpoints in a rolling manner, the technical problem of the difficulty in automatically balancing the objectives of safety, economy and environmental protection at different operation phases is solved, and the technical effect of adaptive adjustment of parameter setpoints with the operation phase is achieved. By using safety verification and full-process coverage logic to send the decision results to the DCS control system, the technical problem of lacking autonomous execution capability without human intervention in the entire process from cold start to shutdown of coal-fired power units has been solved, achieving the technical effect that no human intervention is required in the entire process of decision execution. Attached Figure Description

[0017] Figure 1 This is a flowchart of the fully autonomous decision-making method for multi-process systems in coal-fired power plants provided in this embodiment of the invention. Detailed Implementation

[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0019] like Figure 1 As shown, the fully autonomous decision-making method for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure includes the following steps S100-S400.

[0020] Step S100: Construction of the dual-drive autonomous decision-making algorithm.

[0021] In step S100, based on two driving methods—mechanism model optimization and big data mining—decision rules and parameter settings for each process system of the coal-fired power plant are generated respectively, and the two types of outputs are merged into a unified autonomous decision-making algorithm.

[0022] For example, the mechanism model optimization part is based on the thermodynamic principles of the unit, the combustion mechanism, and the environmental protection reaction mechanism. Optimization models are established for the combustion system, feedwater system, flue gas system, environmental protection system, and auxiliary system, respectively. Each optimization model takes the current operating parameters (including load command, coal quality index, flue gas component concentration, and steam parameters) as input and the optimal control parameter setpoints of each process system as output. Sensor data is collected in real time during unit operation to update the model parameters, so that each optimization model continuously reflects the current thermodynamic state of the unit.

[0023] In at least one example of the embodiments of this disclosure, the big data mining part extracts the historical optimal operating data that meets the preset safety, economic and environmental constraints from the historical operating database. Using the operating condition feature vector (including load range, coal quality classification label, and ambient temperature and humidity) as an index, the historical data is partitioned into operating condition partitions by a clustering algorithm. The input of the clustering algorithm is the operating condition feature vector of the historical operating sample, and the output is the center point of each operating condition partition and the sample set to which it belongs. Within each operating condition partition, the corresponding optimal decision rule is further extracted to form a decision rule library covering different operating conditions.

[0024] It should be noted that the fusion of mechanistic model optimization and big data mining adopts a weighted comprehensive strategy: when historical data covers the current operating conditions, the decision rules output by big data mining are primary, and the output of mechanistic model optimization is secondary; when historical data does not adequately cover the current operating conditions, the output of mechanistic model optimization is primary, and the output of big data mining is secondary. Furthermore, the fusion weight coefficient of the two outputs is calculated in real time based on the distance between the current operating conditions and the center point of historical samples. Specifically, let the feature vector of the current operating conditions be... The Euclidean distance to the nearest historical sample center is The weights output by big data mining Weights of the output of the mechanistic model optimization satisfy: ; in, Weights output for big data mining The weights are used to optimize the output of the mechanistic model. The Euclidean distance between the current feature vector and the center point of the nearest historical sample is given. This is the distance attenuation coefficient, used to control the rate at which weights shift towards the mechanistic model as distance increases. The final parameter setpoints of the fused output. Determined by the following formula: ; in, Set values ​​for the parameters of the final fused output. Assign values ​​to the decision parameters output by big data mining. The decision parameters are set to optimize the output of the mechanism model, thereby compensating for the discrete solution problem of big data mining in the sparse data region and correcting the local optimal deviation of the mechanism model by the historical optimal rule in the dense data region. Finally, the optimal operating mode, equipment scheduling strategy and parameter set values ​​of each process system are automatically determined without human intervention in the decision-making process.

[0025] Step S200: Collaborative decision-making across multiple process systems.

[0026] In step S200, based on the unified autonomous decision-making algorithm generated in step S100, control command sets are generated for the combustion system, water supply system, flue gas system, environmental protection system and auxiliary machine system respectively. The five sets of commands are then subjected to cross-system consistency verification through conflict detection logic. After eliminating control conflicts, collaborative decision-making commands are output.

[0027] For example, for the combustion system, the target values ​​for burner tilt angle, secondary damper opening, and flue gas damper opening are autonomously determined based on the current load command and coal quality indicators to optimize the combustion ratio in the furnace; for the feedwater system, the target values ​​for feedwater flow rate and feedwater temperature are autonomously determined based on the main steam pressure and temperature deviation to achieve precise control of feedwater throughout the process; for the flue gas system, the target values ​​for forced draft fan and induced draft fan speed and corresponding damper opening are autonomously determined based on furnace pressure and flue gas oxygen content to balance the flue gas ratio; for the environmental protection system, the target values ​​for ammonia injection and slurry supply are autonomously determined based on the nitrogen oxide concentration at the denitrification inlet and the sulfur dioxide concentration at the desulfurization inlet; for the auxiliary equipment system, the start-up and shutdown sequence and switching timing of the equipment are autonomously determined based on the operating status and load requirements of each auxiliary machine.

[0028] In at least one example of the embodiments of this disclosure, the conflict detection logic performs the following checks on five sets of instructions: it checks whether the oxygen balance constraint between the combustion ratio instruction of the combustion system and the air volume instruction of the flue gas system is satisfied; it checks whether the flow supply capacity constraint between the water supply flow instruction of the water supply system and the water pump start / stop instruction of the auxiliary system is satisfied; and it checks whether the ammonia-nitrogen molar ratio constraint between the ammonia injection quantity instruction of the environmental protection system and the flue gas flow instruction of the flue gas system is satisfied. When any of the above constraints is not satisfied, the instructions of the relevant systems are iteratively corrected until all constraints are satisfied, and then the final collaborative decision instruction is output.

[0029] Step S300: Autonomous cruise with set operating parameters.

[0030] In step S300, based on a multi-objective optimization algorithm, the system autonomously determines key parameter settings such as main steam pressure, main steam temperature, reheat steam temperature, furnace oxygen content, denitrification setpoint, and desulfurization setpoint according to the current load, coal quality, and operating conditions. The system then adjusts these setpoints in real time during unit operation, taking into account the three objectives of safety, economy, and environmental protection.

[0031] For example, multi-objective optimization algorithms use a three-dimensional objective function vector consisting of a safety objective (maximizing the margin of each parameter deviating from the safety boundary), an economic objective (maximizing the unit's thermal efficiency), and an environmental objective (minimizing the emission concentrations of nitrogen oxides and sulfur dioxide). The decision variable vector consists of main steam pressure, main steam temperature, reheat steam temperature, furnace oxygen content, denitrification setpoint, and desulfurization setpoint. Within the safe operating range of each parameter, the Pareto optimal frontier is solved, and the target weight vector is determined according to the pre-defined target weight vector. Select the parameter set with the highest weighted composite score from the Pareto front. As the current fixed value output. A three-dimensional objective function vector. For security target components, As an economic objective component, As an environmental protection goal, Indicates transpose. For the decision variable vector, For the target weight vector, , , The weights corresponding to the safety, economy, and environmental protection objectives are respectively, satisfying... , The parameter set is the combination that yields the highest weighted composite score.

[0032] In at least one example of the embodiments of this disclosure, the target weight vector Adaptive adjustment based on the current operating phase of the unit: during the variable load peak shaving phase. Improvement; during the steady-state operation phase, Improvement; when environmental emission restriction warnings are triggered, This enhances the system, enabling adaptive cruise control with parameter settings throughout different operating phases without the need for manual configuration.

[0033] It should be noted that there is a parameter consistency constraint between the parameter setpoints output in step S300 and the collaborative decision-making instructions output in step S200: the combustion ratio instructions of the combustion system and the air volume instructions of the flue gas system in step S200 are generated based on the furnace oxygen setpoints output in step S300, and the ammonia injection and slurry supply instructions of the environmental protection system in step S200 are generated based on the denitrification setpoints and desulfurization setpoints output in step S300, respectively, to ensure that the collaborative decision-making instructions and parameter setpoints are compatible at the numerical level.

[0034] Step S400: Autonomous decision-making and execution throughout the entire process.

[0035] In step S400, the collaborative decision-making command output in step S200 and the parameter setpoint output in step S300 are sent to the DCS control system after security verification. The DCS control system automatically completes the execution of control commands, equipment operation and parameter adjustment, covering the entire process of unit cold start, variable load peak shaving, steady-state operation and shutdown.

[0036] For example, safety verification includes the following sub-steps: a) For each collaborative decision instruction and parameter setting, compare it with the preset safety boundary to determine whether it is within the allowable operating range; b) For instructions or settings that exceed the safety boundary, automatically truncate them to the nearest safety boundary value and record the truncation event; c) For auxiliary machine instructions involving equipment start-up and shutdown, verify whether the current equipment status meets the start-up and shutdown preconditions. Instructions that do not meet the preconditions are temporarily suspended and trigger waiting logic; d) When an abnormal operating condition detection signal is triggered, automatically switch to the preset safety decision mode, generate the minimum intervention instruction set to ensure the safe operation of the unit, and issue it with priority.

[0037] In at least one example of the embodiments of this disclosure, full process coverage is achieved in the following manner: In the cold start phase, the mechanism model optimization in step S100 generates start phase parameter setpoints with start-up curve constraints as boundary conditions; in the variable load peak shaving phase, the multi-objective optimization algorithm in step S300 solves in real time with load change rate as an additional constraint; in the steady-state operation phase, the collaborative decision instruction in step S200 is refreshed at a fixed period; in the shutdown phase, the equipment shutdown sequence of the auxiliary system is automatically generated by the auxiliary machine decision logic in step S200 according to the preset shutdown process.

[0038] The fully autonomous decision-making system for a multi-process system of a coal-fired power plant provided in at least one embodiment of this disclosure includes at least one processing node, a decision algorithm construction unit, a collaborative decision-making unit, a setpoint cruise unit, and an autonomous execution unit.

[0039] For a detailed explanation of the decision-making algorithm construction unit, please refer to the relevant description of step S100 in the embodiment of the fully autonomous decision-making method for multi-process systems of coal-fired power plants described above; it will not be repeated here. For a detailed explanation of the collaborative decision-making unit, please refer to the relevant description of step S200 in the embodiment of the fully autonomous decision-making method for multi-process systems of coal-fired power plants described above; it will not be repeated here. For a detailed explanation of the setpoint cruise unit, please refer to the relevant description of step S300 in the embodiment of the fully autonomous decision-making method for multi-process systems of coal-fired power plants described above; it will not be repeated here. For a detailed explanation of the autonomous execution unit, please refer to the relevant description of step S400 in the embodiment of the fully autonomous decision-making method for multi-process systems of coal-fired power plants described above; it will not be repeated here.

[0040] The technical effects of the fully autonomous decision-making system for multi-process systems in coal-fired power plants are the same as those of the aforementioned fully autonomous decision-making method for multi-process systems in coal-fired power plants, and will not be repeated here.

[0041] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A fully autonomous decision-making method for a multi-process system in a coal-fired power plant, characterized in that, include: Based on two driving methods, namely mechanism model optimization and big data mining, decision rules and parameter settings for each process system of coal-fired power plants are generated respectively. The two types of outputs are then fused into a unified autonomous decision-making algorithm through a weighted synthesis strategy. The fusion weight coefficient is calculated in real time based on the Euclidean distance between the current operating condition feature vector and the center point of the historical sample. Based on the unified autonomous decision-making algorithm, control command sets are generated for the combustion system, water supply system, flue gas system, environmental protection system and auxiliary system respectively. The five types of control command sets are then checked for cross-system consistency through conflict detection logic. After eliminating control conflicts, collaborative decision-making commands are output. Based on a multi-objective optimization algorithm, the system autonomously determines the set values ​​of key parameters according to the current load, coal quality, and operating conditions, and adjusts the set values ​​of these key parameters in real time during unit operation. as well as The collaborative decision-making instructions and the key parameter setpoints are sent to the distributed control system after security verification, covering the entire process of unit cold start-up, variable load peak shaving, steady-state operation, and shutdown. The collaborative decision-making instruction is generated based on the set values ​​of the key parameters, and the two satisfy the parameter consistency constraint.

2. The fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to claim 1, characterized in that, The method of fusing the two types of outputs through a weighted synthesis strategy includes: Let the Euclidean distance between the current operating condition feature vector and the center point of the nearest historical sample be . Weights output by big data mining Weights of the output of the mechanistic model optimization satisfy: ； ； in This is the distance attenuation coefficient; The final fused output parameter setpoints Depend on Confirmed, among which Assign values ​​to the decision parameters output by big data mining. The decision parameters output by the mechanistic model optimization are set. Through the above adaptive weight calculation method based on Euclidean distance, the greater the distance between the current working condition and the center point of historical samples, the higher the weight of the mechanistic model optimization, thereby ensuring the reliability of decision-making in working condition intervals where historical data coverage is insufficient.

3. The fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to claim 1, characterized in that, The big data mining process extracts historical operating data that meets the constraints of safety, economy, and environmental protection from the historical operating database. Using the operating condition feature vector as an index, the historical data is partitioned into operating condition zones through a clustering algorithm. The input of the clustering algorithm is the operating condition feature vector of the historical operating sample, and the output is the center point of each operating condition zone and the set of samples to which it belongs. The optimization of the mechanism model is based on the thermodynamic principles of the unit, the combustion mechanism, and the environmental reaction mechanism. Optimization models are established for the combustion system, feedwater system, flue gas system, environmental protection system, and auxiliary system, respectively. Each optimization model takes the current operating parameters as input and the control parameter setpoints of each process system as output.

4. The fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to claim 1, characterized in that, The conflict detection logic performs the following checks on the five sets of control commands: Check whether the oxygen balance constraint between the combustion ratio command of the combustion system and the air volume command of the flue gas system is met; Check whether the flow supply capacity constraint between the water supply flow command of the water supply system and the water pump start / stop command of the auxiliary system is met; Check whether the ammonia-nitrogen molar ratio constraint between the ammonia injection quantity command of the environmental protection system and the flue gas flow command of the flue gas system is satisfied. When any constraint is not satisfied, the instructions of the relevant system are iteratively modified until all constraints are satisfied, and then the collaborative decision instruction is output.

5. The fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to claim 1, characterized in that, The multi-objective optimization algorithm uses safety, economic, and environmental objectives to form a three-dimensional objective function vector. The decision variable vector consists of main steam pressure, main steam temperature, reheat steam temperature, furnace oxygen content, denitrification setpoint, and desulfurization setpoint. The Pareto optimal frontier is solved within the safe operating range of each parameter, and the target weight vector is used as the basis for the solution. Select the parameter set with the highest weighted composite score from the Pareto front. As a fixed value output for the aforementioned key parameter; in, For security target components, As an economic objective component, As an environmental protection goal, , , These are the corresponding weights, satisfying... .

6. The fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to claim 5, characterized in that, The target weight vector Adaptive adjustment based on the current operating stage of the unit: During the load shifting and peak shaving phase promote; During the steady-state operation phase promote; When environmental emission restriction warnings are triggered promote.

7. The fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to claim 1, characterized in that, The security verification includes: For each collaborative decision-making instruction and key parameter setting, compare each item with the preset safety boundary, and truncate any instructions or settings that exceed the safety boundary to the nearest safety boundary value and record the truncation event. For auxiliary machine commands involving equipment start-up and shutdown, check whether the current equipment status meets the preconditions for start-up and shutdown. Commands that do not meet the preconditions are temporarily suspended and a waiting logic is triggered. When an abnormal operating condition detection signal is triggered, the system automatically switches to the preset safety decision mode, generates a minimum set of intervention instructions to ensure the safe operation of the unit, and issues them with priority.

8. A fully autonomous decision-making system for a multi-process system in a coal-fired power plant, characterized in that, A method for implementing a fully autonomous decision-making process for a multi-process system in a coal-fired power plant as described in any one of claims 1 to 7, comprising: The decision algorithm construction unit is configured with two driving methods: mechanism model optimization and big data mining. It generates decision rules and parameter values ​​for each process system of the coal-fired power plant, respectively. The two types of outputs are fused into a unified autonomous decision algorithm through a weighted synthesis strategy. The fusion weight coefficient is calculated in real time based on the Euclidean distance between the current operating condition feature vector and the center point of the historical sample. The collaborative decision-making unit is configured to generate control command sets for the combustion system, water supply system, flue gas system, environmental protection system and auxiliary system respectively according to the unified autonomous decision-making algorithm, and to perform cross-system consistency verification on the five types of control command sets through conflict detection logic, and output collaborative decision-making commands after eliminating control conflicts. The setpoint cruise unit is configured to autonomously determine the setpoints of key parameters based on the current load, coal quality and operating conditions, and adjust the setpoints of key parameters in real time during unit operation, based on a multi-objective optimization algorithm. The autonomous execution unit is configured to send the collaborative decision-making instructions and the key parameter setpoints to the distributed control system after security verification, covering the entire process of unit cold start-up, variable load peak shaving, steady-state operation and shutdown. The collaborative decision-making instruction is generated based on the set values ​​of the key parameters, and the two satisfy the parameter consistency constraint.

9. An electronic device, characterized in that, include: At least one memory stores computer-executable instructions non-transitory; At least one processor, configured to run the computer-executable instructions, The computer-executable instructions are executed by the processor to provide a fully autonomous decision-making method for a multi-process system in a coal-fired power plant according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer-executable instructions, which, when executed by at least one processor, implement the fully autonomous decision-making method for a multi-process system of a coal-fired power plant according to any one of claims 1 to 7.

Images