Method and device for synergistic control of emissions, electronic device and storage medium

Abstract

The present disclosure discloses a synergistic control method and device for emissions, electronic equipment and storage medium. According to the present application, the precise prediction and uncertainty quantification of nitrogen oxide emission concentration are realized through a probability prediction model, and the interpretable emission control rules are extracted and synergistically fused with the preset traditional control strategy to optimize the control logic, thereby improving the control response efficiency under transient operating conditions. Therefore, the technical problem of control instruction lag caused by insufficient calculation efficiency and excessively high load change rate when using model predictive control for transient operating condition control in the prior art can be solved, and the peak breakthrough of nitrogen oxide emission limit value is further triggered. The technical effects of improving the timeliness and accuracy of nitrogen oxide emission control of coal-fired boilers, avoiding emission exceeding the standard, ensuring environmental protection compliance of the unit, and optimizing economic operation indicators are achieved.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and in particular to a method and apparatus for coordinated control of emissions, electronic equipment, and storage medium. Background Technology

[0002] Coal-fired boilers, as core equipment in thermal power generation systems, are widely used in power grid peak shaving and industrial heating. With increasingly stringent environmental regulations on nitrogen oxide (NOx) emissions, existing technologies, through the synergistic operation of model predictive control, deep neural networks, and selective catalytic reduction, have constructed combustion optimization and pollutant control systems.

[0003] However, when using model predictive control directly for transient condition control in existing technologies, the computational efficiency is insufficient, resulting in an excessively high load change rate. This leads to control command lag, causing NOx peaks to exceed emission limits, which affects the unit's environmental compliance and economic operation indicators. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and storage medium for the coordinated control of emissions.

[0005] According to a first aspect of this disclosure, a method for coordinated control of emissions is provided, comprising: Real-time acquisition of multi-dimensional operating parameters during the operation of coal-fired boilers; The multidimensional operating parameters are input into the probabilistic prediction model to predict the nitrogen oxide emission concentration, and the prediction uncertainty measure is obtained. Based on the probabilistic prediction model, interpretable emission control rules are extracted; Based on the predicted uncertainty metric, the interpretable emission control rules, and the preset traditional control strategy, a coordinated control command is generated and executed.

[0006] Optionally, the real-time acquisition of multi-dimensional operating parameters during the operation of the coal-fired boiler includes: Through industrial communication protocols, multiple process parameters related to combustion status, emission status, and unit load are collected from the boiler distributed control system. Transient features are extracted from the collected process parameters to enhance the ability to characterize dynamic changes in load.

[0007] Optionally, the step of inputting the multidimensional operating parameters into the probabilistic prediction model to predict the nitrogen oxide emission concentration and obtaining a prediction uncertainty measure includes: A deep neural network model was used to establish the mapping relationship between operating parameters and nitrogen oxide emission concentration; The deep neural network model is trained based on the variational inference loss function so that its output represents the probability distribution of prediction uncertainty. Noise modeling is performed on the raw data acquired by the sensors to quantify the uncertainty in the measurement process.

[0008] Optionally, extracting interpretable emission control rules based on the probabilistic prediction model includes: Identify key variables that significantly affect the prediction results from the probabilistic prediction model; Based on the symbolic regression algorithm, executable control rules are generated with the key variables as conditions. The control rules are expressions containing mathematical operators.

[0009] Optionally, generating and executing coordinated control instructions based on the prediction uncertainty measure, the interpretable emission control rules, and the preset traditional control strategy includes: Construct a fuzzy arbitration mechanism, the input of which includes the aforementioned prediction uncertainty measure; Based on the magnitude of the prediction uncertainty metric, dynamic weights are assigned to the interpretable emission control rules and the conventional control strategies through the fuzzy arbitration mechanism.

[0010] Optionally, the method further includes: Construct and run multiple synergistic closed-loop control loops; A verification closed loop based on the closed-loop control loop that receives feedback and corrects the control rules; The closed-loop control loop is automatically updated based on the prediction error to update the probability prediction model. A combustion stabilization closed loop is established based on the closed-loop control circuit to monitor the combustion state and ensure the stability of the control actions.

[0011] According to a second aspect of this disclosure, a coordinated emission control device is provided, comprising: The acquisition unit is also used to collect multi-dimensional operating parameters during the operation of coal-fired boilers in real time; The prediction unit is also used to input the multidimensional operating parameters into the probabilistic prediction model to predict the nitrogen oxide emission concentration and obtain a prediction uncertainty measure. The extraction unit is also used to extract interpretable emission control rules based on the probabilistic prediction model; The generation unit is also used to generate and execute coordinated control instructions based on the prediction uncertainty measure, the interpretable emission control rules and the preset traditional control strategy.

[0012] Optionally, the acquisition unit is further configured to: Through industrial communication protocols, multiple process parameters related to combustion status, emission status, and unit load are collected from the boiler distributed control system. Transient features are extracted from the collected process parameters to enhance the ability to characterize dynamic changes in load.

[0013] Optionally, the prediction unit is further configured to: A deep neural network model was used to establish the mapping relationship between operating parameters and nitrogen oxide emission concentration; The deep neural network model is trained based on the variational inference loss function so that its output represents the probability distribution of prediction uncertainty. Noise modeling is performed on the raw data acquired by the sensors to quantify the uncertainty in the measurement process.

[0014] Optionally, the extraction unit is further configured to: Identify key variables that significantly affect the prediction results from the probabilistic prediction model; Based on the symbolic regression algorithm, executable control rules are generated with the key variables as conditions. The control rules are expressions containing mathematical operators.

[0015] Optionally, the generation unit is further configured to: Construct a fuzzy arbitration mechanism, the input of which includes the aforementioned prediction uncertainty measure; Based on the magnitude of the prediction uncertainty metric, dynamic weights are assigned to the interpretable emission control rules and the conventional control strategies through the fuzzy arbitration mechanism.

[0016] Optional, also includes: The arbitration unit is also used to build and run multiple synergistic closed-loop control loops; The correction unit is also used to receive feedback and correct the control rules of the verification closed loop based on the closed-loop control loop; The updating unit is also used to automatically update the model update closed loop of the probability prediction model based on the prediction error of the closed loop control loop. The stabilization unit is also used to monitor the combustion state and ensure the stability of the control action based on the closed-loop control loop.

[0017] According to a third aspect of this disclosure, an electronic device is provided, comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect above.

[0018] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method described in the first aspect above.

[0019] According to a fifth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in the first aspect above.

[0020] The emission coordination control method, apparatus, electronic device, and storage medium disclosed herein, through this application, achieve accurate prediction and uncertainty quantification of nitrogen oxide emission concentrations by using a probabilistic prediction model, while extracting interpretable emission control rules and coordinating and integrating them with preset traditional control strategies to optimize control logic, thereby improving control response efficiency under transient conditions. Therefore, it can solve the technical problems in the prior art where insufficient computational efficiency and excessively high load change rate lead to control command lag when using model predictive control for transient conditions, which in turn causes nitrogen oxide peak emissions to exceed emission limits. This achieves the technical effects of improving the timeliness and accuracy of nitrogen oxide emission control in coal-fired boilers, avoiding emission exceedances, ensuring environmental compliance of the unit, and optimizing economic operation indicators.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0022] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein: Figure 1 A schematic flowchart illustrating a coordinated emission control method provided in an embodiment of this disclosure; Figure 2 A schematic diagram of the structure of a coordinated emission control device provided in an embodiment of this disclosure; Figure 3 A schematic diagram of the structure of a coordinated emission control device provided in an embodiment of this disclosure; Figure 4 A schematic block diagram of an example electronic device provided for embodiments of this disclosure. Detailed Implementation

[0023] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0024] The following description, with reference to the accompanying drawings, describes a method, apparatus, electronic device, and storage medium for coordinated emission control according to embodiments of the present disclosure.

[0025] Figure 1 This is a schematic flowchart illustrating a coordinated emission control method provided in an embodiment of the present disclosure.

[0026] like Figure 1 As shown, the method includes the following steps: Step 101: Real-time acquisition of multi-dimensional operating parameters during the operation of the coal-fired boiler; Through reliable industrial communication and data acquisition mechanisms, multi-dimensional operating parameters of coal-fired boilers are acquired in real time, providing comprehensive, accurate, and timely basic data support for subsequent nitrogen oxide coordinated control. These multi-dimensional operating parameters cover various key parameters related to combustion status, pollutant emissions, and load regulation during the operation of the coal-fired boiler, comprehensively reflecting the boiler's real-time operating conditions.

[0027] Combustion-related parameters include indicators that directly affect combustion completeness, such as furnace temperature field distribution, primary air volume, secondary air volume, and fuel supply rate. Emission-related parameters mainly involve pollutant emission data such as SCR inlet nitrogen oxide concentration. Load-related parameters include parameters reflecting the boiler's output capacity, such as main steam flow rate and unit load value. The data acquisition process uses the OPCUA protocol to achieve data interaction with the distributed control system (DCS). As the basic control and data acquisition system for coal-fired boilers, the DCS can integrate the raw data acquired by sensors distributed in various key parts of the boiler, ensuring the coverage and consistency of parameter acquisition.

[0028] This real-time data acquisition method captures dynamic changes during boiler operation, especially in scenarios with rapid load fluctuations such as peak shaving. Timely acquisition of parameter change information prevents data lag from affecting the adaptability and accuracy of subsequent control strategies, laying a solid data foundation for adaptive optimization control of nitrogen oxides. These acquired multi-dimensional operating parameters are not only core inputs for subsequent feature processing and probabilistic modeling, but also crucial data for ensuring combustion efficiency and system compatibility. The real-time nature and comprehensiveness of their acquisition directly affect the control effectiveness of the entire collaborative control method.

[0029] Step 102: Input the multidimensional operating parameters into the probabilistic prediction model to predict the nitrogen oxide emission concentration and obtain the prediction uncertainty measure; By conducting in-depth analysis of multidimensional operating parameters through a probabilistic prediction model, accurate prediction of nitrogen oxide emission concentrations is achieved, and prediction uncertainty metrics are simultaneously acquired. This provides decision support with both accuracy and reliability for subsequent dynamic coordinated control. Specifically, the probabilistic prediction model employs a Bayesian neural network (BNN). Its core advantage over traditional data-driven models lies in its ability to quantify the uncertainty in the prediction process through probability weights, effectively compensating for the shortcomings of traditional models that only output a single predicted value while ignoring error risks.

[0030] Multidimensional operating parameters, serving as input data for the model, encompass various key indicators related to combustion, emissions, and load. These parameters comprehensively reflect the real-time operating status of the coal-fired boiler, providing a sufficient data foundation for the model to uncover the complex nonlinear mapping relationship between parameters and nitrogen oxide emissions. During model operation, these multidimensional operating parameters are continuously input into the Bayesian neural network. Through the network's hierarchical computation and probabilistic modeling capabilities, the network fully learns the intrinsic correlation between parameter changes and nitrogen oxide emission concentration fluctuations under different operating conditions. The final output includes not only specific nitrogen oxide emission concentration prediction results but also a prediction uncertainty metric that characterizes the reliability of the prediction. This prediction uncertainty metric intuitively reflects the model's confidence in the current prediction results. Its generation stems from the Bayesian neural network's comprehensive consideration of parameter uncertainty, model uncertainty, and data noise, effectively highlighting risk points that need to be addressed in subsequent control strategy formulation.

[0031] By using probabilistic prediction, the accuracy of nitrogen oxide emission concentration prediction is ensured, and uncertainty measurement provides a key reference for the dynamic adjustment of control strategies. This solves the problem of the lack of reliability assessment of traditional model prediction results, and lays an important technical foundation for the adaptive optimization control of nitrogen oxides under the peak-shaving conditions of the unit. At the same time, it provides core input data with confidence labels for subsequent rule extraction and dynamic arbitration.

[0032] Step 103: Based on the probabilistic prediction model, extract interpretable emission control rules; This study extracts interpretable emission control rules directly related to nitrogen oxide emission control from probabilistic prediction models, breaking through the black-box limitations of traditional data-driven models and providing logically clear and directly executable operational guidelines for subsequent coordinated control. The probabilistic prediction model employed is a Bayesian neural network (BNN). In establishing the mapping relationship between multi-dimensional operating parameters and nitrogen oxide emission concentrations, this model can simultaneously quantify the characteristic importance of each input parameter to the prediction results, clearly presenting the degree of influence of different operating parameters on nitrogen oxide generation and emissions, and providing a reliable basis for the accurate selection of key variables.

[0033] By systematically analyzing and ranking the importance of these features, the key variables that have the most significant impact on the predicted nitrogen oxide emission concentration can be screened from multidimensional operating parameters. These key variables cover core dimensions such as combustion state, load change, and pollutant emissions, accurately capturing the core factors affecting nitrogen oxide emissions during boiler operation and laying a targeted foundation for the generation of control rules. After the key variables are determined, a rule generation framework is constructed using a symbolic regression algorithm combined with genetic programming technology. This algorithm can autonomously explore the intrinsic relationships between key variables, integrate various mathematical operators such as addition, subtraction, multiplication, division, and differential operators, and generate executable control rules with a "condition-action" core structure. These rules are presented in the form of intuitive mathematical expressions, clearly indicating the changing trends of the associated key variables and the corresponding control adjustment actions.

[0034] The rule extraction method based on probabilistic prediction models ensures a high degree of alignment between control rules and nitrogen oxide emission patterns. Furthermore, its clear mathematical logic endows the rules with strong interpretability, allowing operators to directly understand the derivation process of control decisions and intervene without relying on specialized algorithmic knowledge. Simultaneously, the generated control rules can accurately adapt to transient operating conditions with rapidly changing loads, such as peak shaving, and respond promptly to dynamic fluctuations in key variables. This provides a logically clear and directly implementable core basis for subsequent dynamic integration with traditional control strategies, further enhancing the operability and adaptability of nitrogen oxide coordinated control.

[0035] Step 104: Generate and execute coordinated control instructions based on the predicted uncertainty measure, the interpretable emission control rules, and the preset traditional control strategy.

[0036] By integrating interpretable emission control rules based on predictive uncertainty metrics with pre-set traditional control strategies through a dynamic collaborative mechanism, scientifically sound collaborative control commands are generated and precisely executed. This achieves adaptive optimization control of nitrogen oxide emissions from coal-fired boilers while ensuring combustion efficiency and stable system operation. Predictive uncertainty metrics serve as a key reference, directly reflecting the reliability of the probabilistic prediction model's output, providing quantitative support for the weight allocation of control strategies, and preventing control errors from causing inaccurate actions. Interpretable emission control rules, based on symbolic regression, provide logically clear operational guidelines, enabling operators to clearly understand the internal logic of control decisions, facilitating intervention and adjustment, and adapting to rapidly changing load conditions.

[0037] The pre-defined traditional control strategy is a mature control scheme that has been proven in practice, providing a fundamental guarantee for coordinated control and ensuring that the system can still operate stably under extreme conditions. The integration of the three is achieved through a dynamic arbitration mechanism, which dynamically allocates the weights of interpretable emission control rules and traditional control strategies based on the specific value of the prediction uncertainty metric. When the prediction confidence is high, the weight of the interpretable rule is increased to give full play to its transient adaptability advantage; when the prediction uncertainty is large, the weight of the traditional control strategy is increased to ensure control stability.

[0038] The generated collaborative control commands must be tailored to the operating characteristics of the coal-fired boiler, balancing nitrogen oxide emission reduction targets with combustion efficiency optimization needs. After generation, the commands are converted into signals recognizable by the actuators via a control command converter, and precisely transmitted to relevant equipment such as burner dampers, ensuring timely implementation of control actions. This entire process overcomes the limitations of traditional static control strategies while mitigating the risks of single control modes through multi-strategy fusion. It effectively adapts to load fluctuation scenarios such as peak shaving, while maintaining good compatibility with existing distributed control systems such as selective catalytic reduction (SCR) and selective non-catalytic reduction (SNR) technologies, facilitating the retrofitting of existing units and the integration of new units.

[0039] In some embodiments, the real-time acquisition of multi-dimensional operating parameters during the operation of the coal-fired boiler includes: Through industrial communication protocols, multiple process parameters related to combustion status, emission status, and unit load are collected from the boiler distributed control system. Transient features are extracted from the collected process parameters to enhance the ability to characterize dynamic changes in load.

[0040] Real-time acquisition of multi-dimensional operating parameters during the operation of a coal-fired boiler is achieved through a mature and stable industrial communication protocol, specifically the OPC UA protocol. This protocol boasts cross-platform compatibility and high reliability, enabling efficient data interaction with the boiler's distributed control system (DCS), ensuring the real-time nature and completeness of parameter acquisition. The acquisition process uses the DCS system as the data source. As the core of basic control and data acquisition for the coal-fired boiler, the DCS integrates sensor data distributed across various key operating components of the boiler. The acquired process parameters comprehensively cover three core dimensions: combustion status, emission status, and unit load.

[0041] Combustion status-related parameters include indicators that directly affect combustion completeness and stability, such as furnace temperature field distribution, primary air volume, secondary air volume, fuel supply rate, and burner operating status. Emission status-related parameters mainly cover key data reflecting pollutant emission levels, such as SCR inlet nitrogen oxide concentration. Unit load-related parameters include parameters that accurately reflect the boiler's output capacity, such as main steam flow, unit load value, and pressure parameters. These parameters together constitute a complete data system reflecting the boiler's real-time operating conditions.

[0042] After completing the process parameter acquisition, it is also necessary to extract transient features from the acquired parameters. By calculating dynamic characteristic indicators such as the first and second derivatives of the parameters, the rate and trend information of parameter changes over time can be extracted. This processing method can effectively enhance the parameter's ability to represent dynamic load changes, solve the problem that the original static parameters are difficult to capture rapid load fluctuations in scenarios such as unit peak shaving, and allow subsequent control strategy formulation to be based on data that is more in line with the dynamic changes of actual operating conditions, providing more targeted input data support for achieving adaptive optimization control of nitrogen oxides.

[0043] In some embodiments, inputting the multidimensional operating parameters into a probabilistic prediction model to predict nitrogen oxide emission concentration and obtaining a prediction uncertainty measure includes: A deep neural network model was used to establish the mapping relationship between operating parameters and nitrogen oxide emission concentration; The deep neural network model is trained based on the variational inference loss function so that its output represents the probability distribution of prediction uncertainty. Noise modeling is performed on the raw data acquired by the sensors to quantify the uncertainty in the measurement process.

[0044] The deep neural network model used is a Bayesian neural network (BNN). This model can deeply explore the complex nonlinear mapping relationship between multi-dimensional operating parameters and nitrogen oxide emission concentration by constructing a multi-layer network structure. The multi-dimensional operating parameters, which include combustion state, emission state, and unit load-related data, provide the model with rich learning samples, enabling it to fully adapt to the parameter variation patterns under different operating conditions of coal-fired boilers.

[0045] During model training, the variational inference loss function, also known as the evidence lower bound (ELBO), is used as the optimization objective. By continuously iterating and adjusting the model's probability weight parameters, the model gradually acquires the ability to output a probability distribution. This probability distribution not only includes the predicted value of nitrogen oxide emission concentration but also accurately characterizes the degree of uncertainty in the prediction results, effectively solving the problem that traditional models only output a single predicted value and cannot assess error risk. Simultaneously, to ensure the comprehensiveness and accuracy of the prediction uncertainty measurement, noise modeling of the raw data collected by the sensors is also required. Specifically, the Gaussian process modeling method is used to quantify and analyze the random noise generated during sensor operation, clarifying the impact of noise on the accuracy of the raw data and incorporating the uncertainty of the measurement process into the overall prediction uncertainty measurement system.

[0046] By combining the model's own predictive uncertainty with the data measurement uncertainty, the final predictive uncertainty metric is more in line with the actual situation, providing a more reliable quantitative basis for the formulation of subsequent control strategies and weight allocation, and further improving the accuracy and adaptability of nitrogen oxide coordinated control.

[0047] In some embodiments, extracting interpretable emission control rules based on the probabilistic prediction model includes: Identify key variables that significantly affect the prediction results from the probabilistic prediction model; Based on the symbolic regression algorithm, executable control rules are generated with the key variables as conditions. The control rules are expressions containing mathematical operators.

[0048] The probabilistic prediction model employed is a Bayesian Neural Network (BNN). This model, in establishing the mapping relationship between multidimensional operating parameters and nitrogen oxide emission concentrations, can simultaneously output a measure of the feature importance of each input variable to the prediction results, providing a quantitative basis for identifying key variables. By analyzing and ranking the importance of these features, key variables that significantly affect the predicted nitrogen oxide emission concentrations can be accurately identified. These key variables cover core parameters related to combustion state load changes and emissions, directly reflecting the core factors affecting nitrogen oxide generation and emissions during boiler operation, laying a targeted foundation for the subsequent generation of control rules. After identifying the key variables, a symbolic regression algorithm combined with genetic programming techniques is used to construct a rule generation framework with the key variables as the core conditions.

[0049] This algorithm autonomously explores the inherent mathematical relationships between key variables. By integrating various mathematical operators such as addition, subtraction, multiplication, division, and differentiation, it generates executable control rules with clear logical relationships. These control rules are presented in the form of expressions containing mathematical operators, forming a clear "condition-action" structure that directly links changes in key variables with corresponding control adjustment actions. This rule generation method based on key variables and symbolic regression ensures a strong correlation between control rules and nitrogen oxide emission prediction results, while also providing good interpretability through intuitive mathematical expressions. Operators can clearly understand the derivation logic of control decisions, effectively solving the problems of difficult-to-understand decision logic and unclear operational guidance caused by the black-box nature of traditional data-driven models. Simultaneously, the generated control rules can accurately adapt to the dynamic changes of key variables, especially under conditions of rapid load fluctuations such as unit peak shaving, providing timely and targeted operational guidance. This provides a core basis with clear execution logic for subsequent dynamic arbitration integration with traditional control strategies, further ensuring the adaptability and operability of nitrogen oxide collaborative control.

[0050] In some embodiments, generating and executing coordinated control instructions based on the prediction uncertainty measure, the interpretable emission control rules, and a preset conventional control strategy includes: Construct a fuzzy arbitration mechanism, the input of which includes the aforementioned prediction uncertainty measure; Based on the magnitude of the prediction uncertainty metric, dynamic weights are assigned to the interpretable emission control rules and the conventional control strategies through the fuzzy arbitration mechanism.

[0051] According to the method described in claim 1, a fuzzy arbitration mechanism needs to be constructed first during the process of generating and executing collaborative control instructions. The core function of this mechanism is to realize the dynamic integration of interpretable emission control rules and preset traditional control strategies, solve the problem of insufficient adaptability of a single control strategy under different operating conditions, and ensure the scientific nature and stability of collaborative control.

[0052] The key input to the fuzzy arbitration mechanism is the prediction uncertainty metric, which directly reflects the reliability of the probabilistic prediction model's output. Its value quantifies the confidence level of the model's predictions under the current operating conditions, providing an objective and accurate quantitative basis for subsequent weight allocation. During weight allocation, the fuzzy arbitration mechanism dynamically adjusts the weights based on the specific value of the prediction uncertainty metric, forming a strategy fusion ratio adapted to the current operating conditions.

[0053] When the prediction uncertainty is small, it means that the probability prediction model has high reliability in predicting nitrogen oxide emission concentrations. In this case, the fuzzy arbitration mechanism will increase the weight of interpretable emission control rules, giving full play to its advantages in targeted operational guidance based on key variables, adapting to transient operating conditions with rapid load changes such as unit peak shaving, and achieving precise control of nitrogen oxide emissions. When the prediction uncertainty is large, it indicates that the model prediction results are greatly affected by factors such as operating condition fluctuations and data noise, resulting in reduced reliability. In this case, the fuzzy arbitration mechanism will correspondingly increase the weight of the preset traditional control strategy, relying on the mature logic of the traditional control strategy that has been verified in practice, to ensure the stability of the combustion process of coal-fired boilers and avoid control inaccuracies due to prediction errors.

[0054] This dynamic weight allocation is not a fixed-proportion switch, but a smooth transition based on the continuous changes in the measurement of prediction uncertainty by a fuzzy arbitration mechanism. This ensures that the integration of the two control strategies is natural and without abrupt changes. It fully leverages the transient adaptability of interpretable emission control rules while using the stability of traditional control strategies as a safety net for the system. This allows the generated collaborative control commands to take into account nitrogen oxide emission reduction targets, combustion efficiency optimization, and system operational stability, while maintaining good compatibility with existing distributed control systems. It is suitable for both retrofitting existing units and integrating new units.

[0055] In some embodiments, the method further includes: Construct and run multiple synergistic closed-loop control loops; A verification closed loop based on the closed-loop control loop that receives feedback and corrects the control rules; The closed-loop control loop is automatically updated based on the prediction error to update the probability prediction model. A combustion stabilization closed loop is established based on the closed-loop control circuit to monitor the combustion state and ensure the stability of the control actions.

[0056] As a crucial component, the verification closed loop continuously receives various feedback information from the actual operation process, relying on the closed-loop control circuit. It focuses on collecting feedback from operators regarding the effectiveness of control rule execution. This feedback directly reflects the degree of adaptation of the control rules to actual operating conditions. By analyzing and evaluating the feedback data, the verification closed loop identifies deviations or deficiencies in the control rules and then precisely corrects them, enabling the control rules to better fit the actual operating characteristics of the coal-fired boiler.

[0057] The model update closed loop is based on the prediction error of the probabilistic prediction model. The closed-loop control loop monitors the deviation between the predicted results output by the probabilistic prediction model and the actual nitrogen oxide emission concentration in real time. When the prediction error exceeds the preset reasonable range, the model update closed loop automatically triggers the update process of the probabilistic prediction model. By retraining and adjusting the model parameters, the model structure is optimized, continuously improving the model's adaptability to complex operating conditions and prediction accuracy, ensuring that the model always maintains good prediction performance. The combustion stability closed loop focuses on monitoring the combustion state. Through the closed-loop control loop, it integrates sensor data distributed in key parts of the boiler to capture key information reflecting combustion stability in real time, such as the furnace flame state, furnace temperature field distribution, and burner operating status. It dynamically tracks and evaluates changes in the combustion state after the control action is implemented. If fluctuations or instability in the combustion state are detected, the combustion stability closed loop will promptly issue adjustment signals to correct the control action, avoiding combustion imbalance caused by control strategy adjustments, and ensuring that the boiler maintains a highly efficient and stable combustion state while achieving the nitrogen oxide emission reduction target.

[0058] The three closed-loop control loops do not operate independently but work together. The rule correction of the verification loop provides a more realistic training basis for the model to update the closed loop, and the state monitoring of the combustion stability closed loop defines the safety boundary for the adjustment of the first two closed loops. The three form an organic whole that supports and constrains each other, which significantly improves the robustness and adaptability of the entire control method. This ensures that the control strategy can be continuously optimized and always maintain a safe and stable operating state under complex operating conditions such as rapid load changes such as unit peak shaving.

[0059] Corresponding to the aforementioned coordinated emission control method, this invention also proposes a coordinated emission control device. Since the device embodiments of this invention correspond to the aforementioned method embodiments, details not disclosed in the device embodiments can be referred to the aforementioned method embodiments, and will not be repeated here.

[0060] Figure 2 This is a schematic diagram of the structure of a coordinated emission control device provided in an embodiment of the present disclosure, as shown below. Figure 2 As shown, it includes: The acquisition unit 21 is also used to acquire multi-dimensional operating parameters during the operation of the coal-fired boiler in real time; Prediction unit 22 is also used to input the multidimensional operating parameters into the probabilistic prediction model to predict the nitrogen oxide emission concentration and obtain a prediction uncertainty measure; Extraction unit 23 is also used to extract interpretable emission control rules based on the probability prediction model; The generation unit 24 is also used to generate and execute coordinated control instructions based on the prediction uncertainty measure, the interpretable emission control rules and the preset traditional control strategy.

[0061] Furthermore, in one possible implementation of the embodiments of this disclosure, such as Figure 3 As shown, the acquisition unit 21 is also used for: Through industrial communication protocols, multiple process parameters related to combustion status, emission status, and unit load are collected from the boiler distributed control system. Transient features are extracted from the collected process parameters to enhance the ability to characterize dynamic changes in load.

[0062] Furthermore, in one possible implementation of the embodiments of this disclosure, such as Figure 3 As shown, the prediction unit 22 is further used for: A deep neural network model was used to establish the mapping relationship between operating parameters and nitrogen oxide emission concentration; The deep neural network model is trained based on the variational inference loss function so that its output represents the probability distribution of prediction uncertainty. Noise modeling is performed on the raw data acquired by the sensors to quantify the uncertainty in the measurement process.

[0063] Furthermore, in one possible implementation of the embodiments of this disclosure, such as Figure 3 As shown, the extraction unit 23 is further used for: Identify key variables that significantly affect the prediction results from the probabilistic prediction model; Based on the symbolic regression algorithm, executable control rules are generated with the key variables as conditions. The control rules are expressions containing mathematical operators.

[0064] Furthermore, in one possible implementation of the embodiments of this disclosure, such as Figure 3 As shown, the generation unit 24 is further configured to: Construct a fuzzy arbitration mechanism, the input of which includes the aforementioned prediction uncertainty measure; Based on the magnitude of the prediction uncertainty metric, dynamic weights are assigned to the interpretable emission control rules and the conventional control strategies through the fuzzy arbitration mechanism.

[0065] Furthermore, in one possible implementation of the embodiments of this disclosure, such as Figure 3 As shown, it also includes: Arbitration unit 25 is also used to construct and operate multiple synergistic closed-loop control loops; The correction unit 26 is also used to receive feedback and correct the control rules based on the closed-loop control loop for verification closed loop; The updating unit 27 is also used to automatically update the model update closed loop of the probability prediction model based on the prediction error of the closed loop control loop. The stabilization unit 28 is also used to monitor the combustion state and ensure the stability of the control action based on the closed-loop control loop.

[0066] It should be noted that the foregoing explanation of the method embodiments also applies to the apparatus of the embodiments of this disclosure, and the principle is the same. Therefore, the embodiments of this disclosure are not limited thereto.

[0067] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0068] Figure 4 A schematic block diagram of an example electronic device 400 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0069] like Figure 4As shown, device 400 includes a computing unit 401, which can perform various appropriate actions and processes based on a computer program stored in ROM (Read-Only Memory) 402 or a computer program loaded from storage unit 408 into RAM (Random Access Memory) 403. RAM 403 may also store various programs and data required for the operation of device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via bus 404. I / O (Input / Output) interface 405 is also connected to bus 404.

[0070] Multiple components in device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of monitors, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0071] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, CPUs (Central Processing Units), GPUs (Graphics Processing Units), various special-purpose AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, DSPs (Digital Signal Processors), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as the coordinated emission control method. For example, in some embodiments, the coordinated emission control method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of the methods described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform the aforementioned coordinated control method for emissions by any other suitable means (e.g., by means of firmware).

[0072] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application-Specific Standard Products), SOCs (System-on-Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0073] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0074] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, RAM, ROM, EPROM (Electrically Programmable Read-Only Memory) or flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0075] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0076] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include LANs (Local Area Networks), WANs (Wide Area Networks), the Internet, and blockchain networks.

[0077] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service system that addresses the shortcomings of traditional physical hosts and VPS (Virtual Private Server) services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0078] It's important to note that artificial intelligence (AI) is the study of enabling computers to simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). It encompasses both hardware and software technologies. AI hardware technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, and big data processing. AI software technologies primarily include computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graph technologies.

[0079] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0080] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for coordinated emission control, characterized in that, include: Real-time acquisition of multi-dimensional operating parameters during the operation of coal-fired boilers; The multidimensional operating parameters are input into the probabilistic prediction model to predict the nitrogen oxide emission concentration, and the prediction uncertainty measure is obtained. Based on the probabilistic prediction model, interpretable emission control rules are extracted; Based on the predicted uncertainty metric, the interpretable emission control rules, and the preset traditional control strategy, a coordinated control command is generated and executed.

2. The method according to claim 1, characterized in that, The real-time acquisition of multi-dimensional operating parameters during the operation of the coal-fired boiler includes: Through industrial communication protocols, multiple process parameters related to combustion status, emission status, and unit load are collected from the boiler distributed control system. Transient features are extracted from the collected process parameters to enhance the ability to characterize dynamic changes in load.

3. The method according to claim 1, characterized in that, The step of inputting the multidimensional operating parameters into the probabilistic prediction model to predict nitrogen oxide emission concentration and obtaining a prediction uncertainty measure includes: A deep neural network model was used to establish the mapping relationship between operating parameters and nitrogen oxide emission concentration; The deep neural network model is trained based on the variational inference loss function so that its output represents the probability distribution of prediction uncertainty. Noise modeling is performed on the raw data acquired by the sensors to quantify the uncertainty in the measurement process.

4. The method according to claim 1, characterized in that, The extraction of interpretable emission control rules based on the probabilistic prediction model includes: Identify key variables that significantly affect the prediction results from the probabilistic prediction model; Based on the symbolic regression algorithm, executable control rules are generated with the key variables as conditions. The control rules are expressions containing mathematical operators.

5. The method according to claim 1, characterized in that, The step of generating and executing coordinated control instructions based on the prediction uncertainty measure, the interpretable emission control rules, and the preset traditional control strategy includes: Construct a fuzzy arbitration mechanism, the input of which includes the aforementioned prediction uncertainty measure; Based on the magnitude of the prediction uncertainty metric, dynamic weights are assigned to the interpretable emission control rules and the conventional control strategies through the fuzzy arbitration mechanism.

6. The method according to claim 1, characterized in that, The method further includes: Construct and run multiple synergistic closed-loop control loops; A verification closed loop based on the closed-loop control loop that receives feedback and corrects the control rules; The closed-loop control loop is automatically updated based on the prediction error to update the probability prediction model. A combustion stabilization closed loop is established based on the closed-loop control circuit to monitor the combustion state and ensure the stability of the control actions.

7. A coordinated emission control device, characterized in that, include: The acquisition unit is also used to collect multi-dimensional operating parameters during the operation of coal-fired boilers in real time; The prediction unit is also used to input the multidimensional operating parameters into the probabilistic prediction model to predict the nitrogen oxide emission concentration and obtain a prediction uncertainty measure. The extraction unit is also used to extract interpretable emission control rules based on the probabilistic prediction model; The generation unit is also used to generate and execute coordinated control instructions based on the prediction uncertainty measure, the interpretable emission control rules and the preset traditional control strategy.

8. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-6.

10. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method according to any one of claims 1-6.

Images