Intelligent denitration system and treatment method

Abstract

The invention discloses an intelligent denitration system and a processing method, and relates to the field of denitration processing, and the method comprises the following steps: constructing a distributed control system based on Modbus communication, opening a Modbus communication interface, and carrying out data acquisition through an acquisition device, the acquired data being first acquisition data; creating a denitration artificial intelligence model; creating a strategy model, and cooperatively working with the denitration artificial intelligence model; an execution signal is generated, the execution signal is sent to execution equipment, the ammonia water feeding amount is adjusted through the execution equipment, and a PID controller is adopted for adjustment in the adjustment process; a closed-loop control system is established, and a main PID controller and an auxiliary PID controller are arranged. According to the application, data acquisition is carried out through the acquisition device, acquisition of related data such as the boiler, the denitration system and the discharge port concentration is realized, and accurate denitration effect prediction data is generated through creation of the denitration artificial intelligence model and the strategy model and cooperative work of the denitration artificial intelligence model and the strategy model.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of denitrification treatment, and in particular to an intelligent denitrification system and treatment method. Background Art

[0002] In the fluidized bed boiler industry, due to the complexity and variability of the combustion process in the furnace, a large amount of nitrogen oxides will be produced. These nitrogen oxides are an important cause of acid rain and photochemical smog, which will cause serious damage to air quality and directly affect the human respiratory system. Therefore, denitrification systems are currently commonly used to reduce nitrogen oxide emissions. In existing technologies, denitrification systems usually include key components such as ammonia injection devices, catalyst reactors, and flue gas recirculation systems. By precisely controlling the ammonia injection amount and reaction temperature, the efficient conversion of nitrogen oxides into harmless substances is ensured. At the same time, emission data is monitored in real time to ensure stable operation of the system and meet environmental protection standards.

[0003] However, existing denitrification systems still face several challenges. For example, due to delayed DCS control and regulation responses, CEMS data lags, and the inability of current ammonia pipeline control valves to meet online adjustment requirements, current operational monitoring relies solely on manual adjustments of ammonia flow and ozone generator power based on changes in nitrogen oxide concentration at the total outlet, making monitoring difficult. Furthermore, to ensure compliance with environmental standards, nitrogen oxide emissions are kept at a low level, resulting in high ammonia and ozone generator power consumption. Therefore, further improvements are needed. Summary of the Invention

[0004] The embodiments of the present application provide an intelligent denitrification system and treatment method for effectively balancing the optimal control and economic operation of the denitrification system.

[0005] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions: In a first aspect, a treatment method for a smart denitrification system is provided, the method comprising: Constructing a distributed control system based on Modbus communication, opening a Modbus communication interface, and collecting data through an acquisition device, wherein the collected data is first collected data, and constructing a data link for the first collected data; Creating a denitrification artificial intelligence model, the denitrification artificial intelligence model having an input end and an output end, the input end being used to input raw data, and the output end being used to output denitrification effect prediction data processed by the model; Create a strategy model that includes an outlet NOx concentration prediction model and works in conjunction with the denitrification artificial intelligence model; Generate an execution signal and send it to the execution device, which adjusts the amount of ammonia solution added, using a PID controller for regulation during the adjustment process; Establish a closed-loop control system and set up a main PID controller and a sub-PID controller. The main PID controller is responsible for calculating the inlet NOx concentration value, the set value, and the outlet NOx concentration value, while the sub-PID controller is responsible for executing the equipment adjustment. A big data model library is established to store routine operation data and adjustment methods into the memory library, including inlet flue gas temperature, flow, initial NOx concentration and distribution, ammonia injection amount, intelligent valve adjustment status, outlet NOx concentration, outlet flue gas temperature, oxygen content, and enhance the scale of the model to make subsequent operations more optimized and smarter, thus having self-learning capabilities.

[0006] In another possible implementation of the first aspect, establishing the data link includes: Use Modbus to read data from the distributed control system and then develop the Modbus interface; Conduct multiple rounds of stress testing on the interface software, simulating high-concurrency scenarios to verify its load-bearing capacity; The interface software is called by the first device, and data protocol conversion is performed after the data is read, and the called data is pre-processed, and then the pre-processed data is stored in the first device, where the first device is an RK3199 single-board computer; Create data storage and query structures and link them with system clients. System clients query real-time data and historical trends by selecting interface programs and database storage directories.

[0007] In another possible implementation manner of the first aspect, after the data protocol conversion is performed, a preprocessing operation is required, and the preprocessing operation includes data noise reduction processing and 0-2 interpolation processing.

[0008] In another possible implementation of the first aspect, creating the denitrification artificial intelligence model further includes: A model input layer is set to receive raw data of the denitrification process, where the raw data is first collected data stored in the distributed control system and collected by the collection device; Set up the model's hidden layer and receive data from the model's input layer. The model's hidden layer uses a multi-layer neural network structure to extract features from the data layer by layer. The number of layers and nodes in the multi-layer neural network structure needs to be dynamically adjusted according to the actual data characteristics. Set up the model training and validation layers, create training sets and validation sets, perform feature normalization, then train the model using the training set and validate the model using the validation set.

[0009] In another possible implementation of the first aspect, in the training and validation layers, the training and validation processes are placed at the same level, and the deviation is adjusted during the training and validation processes.

[0010] In another possible implementation of the first aspect, the outlet nitrogen oxide concentration prediction model has an input end, which is used to input the first collected data and the features extracted by the model hidden layer, and an output end, which is used to output the predicted outlet nitrogen oxide concentration, and the nitrogen oxide controlled variable is generated by comparing and fitting the concentration with the actual collected outlet nitrogen oxide concentration.

[0011] In another possible implementation of the first aspect, an execution signal is generated and sent to an execution device, and the execution signal is linked with the execution device through the execution signal, wherein the execution device is an intelligent pneumatic control valve or a combination of an intelligent pneumatic control valve and an intelligent electric control valve.

[0012] In a second aspect, the present application provides a smart denitrification system, which includes: A distributed control system construction module is used to construct a distributed control system based on Modbus communication, open a Modbus communication interface, and collect data through an acquisition device. The collected data is the first collected data, and a data link is established for the first collected data. A denitrification artificial intelligence model creation module is used to create a denitrification artificial intelligence model. The denitrification artificial intelligence model has an input end and an output end. The input end is used to input raw data, and the output end is used to output denitrification effect prediction data processed by the model. A strategy model creation module is used to create a strategy model, which contains an outlet nitrogen oxide concentration prediction model and works in conjunction with the denitrification artificial intelligence model; An execution module is used to generate an execution signal and send the execution signal to an execution device, and the ammonia dosage is adjusted by the execution device, and a PID controller is used for regulation during the adjustment process; Closed-loop control module, used to establish a closed-loop control system and set up a main PID controller and a sub-PID controller. The main control PID controller is responsible for calculating the inlet NOx concentration value, the set value and the outlet NOx concentration value, and the sub-PID controller is responsible for executing the equipment adjustment. The big data model library establishment module is used to establish a big data model library, storing the normal operation data and adjustment methods into the memory library, including inlet flue gas temperature, flow, NOx initial concentration and distribution, ammonia injection amount, intelligent valve adjustment status, outlet NOx concentration, outlet flue gas temperature, oxygen content, and enhancing the scale of the model to make subsequent operation more optimized and smarter, thus having self-learning function.

[0013] Through the above technical solution, by building a distributed control system based on Modbus communication, opening the Modbus communication interface, and collecting data through acquisition equipment, the collection of relevant data such as boilers, denitrification systems, and outlet concentrations is realized. By creating a denitrification artificial intelligence model and a strategy model, the two work together to generate accurate denitrification effect prediction data, and combine with the PID controller for fine adjustment to achieve precise control of the ammonia dosage, ensuring that the outlet nitrogen oxide concentration meets the standard. Through the closed-loop control system, by setting the main PID controller and the sub-PID controller, the ammonia dosage is dynamically optimized, which not only ensures the denitrification effect, but also effectively reduces the operating cost, maximizes resource utilization, improves the overall system stability, and ensures long-term efficient operation.

[0014] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a process flow of a treatment method for a smart denitrification system provided in an embodiment of the present application; Figure 2 Schematic diagram of the module structure of the intelligent denitrification system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0017] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), such directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0018] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0019] Example 1: Figure 1 The following schematically shows a flow chart of the treatment method of the smart denitrification system according to an embodiment of the present application. Figure 1 As shown, the embodiment of the present application provides a treatment method of a smart denitrification system, which is applied to the process of removing nitrogen oxides from combustion flue gas, thereby reducing the direct emission of nitrogen oxides into the air and causing air pollution. The method includes the following steps: S110: Build a distributed control system based on Modbus communication, open a Modbus communication interface, and collect data through an acquisition device, where the collected data is first collected data, and establish a data link for the first collected data; A distributed control system (DCS) is a computer control system that centrally manages and decentralized controls production processes. Modbus is a serial communication protocol commonly used for data exchange between industrial automation devices. Common Modbus modes include RTU and TCP. In a distributed control system, multiple devices are located in different locations, so the mode used is not limited in this embodiment. The distributed control system includes the sensor / actuator layer, PLC / RTU layer, SCADA layer, and upper computer management layer. Devices at each layer need to be connected via Modbus. For example, the sensor sends data to the PLC via the RTU protocol, and the PLC then uploads the data to the SCADA system via Ethernet, or directly opens interfaces for other systems to call, thereby realizing data collection, analysis, processing, execution, and other processes, and reading relevant data such as boilers, denitrification systems, and exhaust concentrations. The data collection equipment includes but is not limited to analyzers, temperature sensors, pressure sensors, flow sensors, etc. These sensors are responsible for real-time monitoring of various parameters in the flue gas and transmitting the data to the PLC / RTU layer for processing. Moreover, through the distributed control system based on Modbus communication, the data parameters can be transmitted to the existing control system via the Modbus RTU protocol, and free switching between DCS control and intelligent control can be achieved through non-disruptive switching configuration; After the first data is collected, it will be temporarily stored in the distributed control system. In order to ensure the real-time and accuracy of the data, a data link needs to be established. The data link construction includes: S210, use Modbus to read data from the distributed control system, and then develop the Modbus interface; The stability of the Modbus interface software directly affects the stability of the entire system. The normal operation of the system has high requirements for the real-time performance of the data, so the stability of the interface software is extremely high. Therefore, after reading the data, it is necessary to develop the Modbus interface and perform stability testing on the interface software to ensure that the data transmission is correct and meets the real-time requirements of the system. S220: Perform multiple rounds of stress testing on the interface software, simulating high-concurrency scenarios to verify its load-bearing capacity; In order to ensure the stability of the interface software under high load, multiple rounds of stress testing are required to simulate high concurrency situations in actual operation, to ensure that it can still transmit data stably under extreme conditions and respond to system instructions in a timely manner to avoid data loss or delay, thereby ensuring the reliability and efficiency of the entire control system. Among them, the multiple rounds of stress testing of the interface software include but are not limited to simulating scenarios such as simultaneous requests from different devices and peak data transmission. For details, please refer to the existing technology and will not be described in detail in this embodiment; S230: Calling the interface software through the first device, performing data protocol conversion after reading the data, preprocessing the called data, and then storing the preprocessed data in the first device; During the entire process, a large amount of data collection and transmission processes need to be generated. Since the data formats of different devices are different, data protocol conversion is required to ensure seamless data connection between different devices. In this embodiment, the interface software is called by the first device, which is the RK3199 single-board computer. The RK3199 single-board computer has powerful data processing capabilities and can efficiently complete data protocol conversion and preprocessing tasks; After data protocol conversion, preprocessing is required. This includes data noise reduction and 0-2 interpolation to eliminate noise interference in the data and fill in possible data gaps to ensure data integrity and accuracy. Finally, the processed high-quality data is stored in the RK3199 single-board computer to provide a basis for subsequent data analysis and control decisions. It should be noted that the entire data transmission process is implemented using interface software to ensure that data is not tampered with or lost during transmission. At the same time, the interface software also has a self-monitoring function that can monitor the data transmission status in real time. Once an abnormality is found, it will immediately issue an early warning and suspend transmission to ensure data security and system stability. At the same time, the single-board computer RK3199 is not connected to the external network during operation, and the underlying data of the distributed control system is directly collected through the interface software, further ensuring the closedness and security of data transmission, and effectively preventing external network attacks and data leakage risks; S240: Creates a data storage and query structure and links it with the system client. The system client can query real-time data and historical trends by selecting the interface program and database storage directory. When querying historical trends, any trend group has multiple measurement points. Users can select specific measurement points for detailed analysis as needed. The system will automatically retrieve the corresponding data and generate visual charts to intuitively display data change trends. It also supports export functions to facilitate further data mining and application. S120: Create a denitrification artificial intelligence model, wherein the denitrification artificial intelligence model has an input end and an output end, wherein the input end is used to input raw data, and the output end is used to output denitrification effect prediction data processed by the model; In denitrification treatment, combining artificial intelligence with data analysis and control can reduce blind spots in production process measurement, change the situation in existing technologies where original parameters lag and specific data cannot be detected, and timely adjust parameters to generate control strategies to ensure smooth production transitions during changes in load and pollutant concentrations, reduce nitrogen oxide emissions, and improve environmental benefits. The model uses a deep learning algorithm to continuously optimize model parameters and improve prediction accuracy through training with a large amount of historical data. Specifically, creating a denitrification artificial intelligence model further includes the following steps: S220: Setting a model input layer to receive raw data of the denitrification process, where the raw data is first collected data stored in the distributed control system and collected by a collection device; Setting up an input layer for the model can effectively capture and process multi-dimensional data, ensuring the comprehensiveness and accuracy of the model input, thereby improving the model's prediction accuracy for the denitrification effect. The model input layer receives first collected data, which includes key parameters such as temperature, pressure, and flow rate. The first collected data is standardized through data preprocessing. The standardization process includes steps such as data cleaning, data denoising, and normalization. For details, please refer to existing technologies. S320, setting a model hidden layer and receiving data transmitted from the model input layer. The model hidden layer adopts a multi-layer neural network structure to extract features from the data layer by layer; A hidden layer is set for the model, and the preprocessed data in the above process is input into the model hidden layer. The model hidden layer extracts key features layer by layer through a multi-layer neural network, enhancing the model's ability to capture complex relationships, further optimizing the model's prediction accuracy for denitrification effects, and ensuring the efficiency and accuracy of data processing. The number of layers and nodes in the multi-layer neural network structure needs to be dynamically adjusted according to the actual data characteristics. For example, a 3-layer or 5-layer network can be selected based on the data complexity, with the number of nodes per layer ranging from 64 to 128 to achieve the optimal feature extraction effect; S420: Set the model training and validation layers, create a training set and a validation set, perform feature normalization, and then train the model using the training set and validate the model using the validation set. Artificial intelligence models usually involve training sets and validation sets. The training set is used to train the model in combination with the extracted feature data, and the validation set is used to evaluate the model performance and ensure its stability and accuracy in practical applications. In this embodiment, by setting up training and validation layers, the two processes are placed on the same level, which facilitates real-time monitoring of the model training progress and validation results, timely detection and adjustment of deviations in training, and improved model training efficiency. S130. Create a strategy model, which includes an outlet nitrogen oxide concentration prediction model and works in conjunction with a denitrification artificial intelligence model; Among them, the outlet nitrogen oxide concentration prediction model has multiple input terminals, which are used to input the first collected data and the features extracted by the model, and an output terminal, which is used to output the predicted outlet nitrogen oxide concentration. The concentration is compared and fitted with the actual collected outlet nitrogen oxide concentration to generate a nitrogen oxide controlled variable. Specifically, the comparison and fitting adopts the least squares method to calculate the error between the predicted value and the actual value, and continuously optimize the model parameters until the error is minimized to ensure the prediction accuracy. For example, the predicted outlet nitrogen oxide concentration is 65mg / Nm³, and the actual collected value is 68mg / Nm³, with an error of 3mg / Nm³. When the least squares method is used for comparison and fitting calculation, the sum of squares of the error between the required data and the actual data can be minimized, and the data change trend of the aforementioned process is fitted, and the model parameters are finally determined to make the predicted value closer to the actual value; S140, generating an execution signal and sending the execution signal to an execution device, and adjusting the amount of ammonia solution added by the execution device, using a PID controller for regulation during the adjustment process; After model analysis, the amount of ammonia to be added is generated. To feed the calculated results into actual use, an execution signal needs to be generated and linked to the execution device. The execution device is an intelligent pneumatic control valve. By adjusting the valve opening of the intelligent pneumatic control valve, the ammonia flow rate is precisely controlled to ensure that it matches the predicted nitrogen oxide concentration and maximize the denitrification efficiency. At the same time, the valve opening and ammonia flow rate are monitored in real time, and the PID parameters are dynamically adjusted to maintain stable system operation. In addition to intelligent pneumatic control valves, common intelligent electric control valves can also be combined. By combining pneumatic and electric control valves, control accuracy can be further improved, modification costs can be reduced, and more flexible adjustment strategies can be implemented to ensure efficient denitrification under different working conditions. The PID controller is a commonly used feedback control algorithm that achieves precise control of the system through the combination of three parameters: proportional (P), integral (I), and differential (D). The initial parameters and control deviation acquisition method of the PID controller can refer to the existing technology and will not be described in detail in this embodiment.

[0020] S150, establish a closed-loop control system and set a main PID controller and a sub-PID controller, wherein the main control PID controller is responsible for calculating the inlet nitrogen oxide concentration value, the set value and the outlet nitrogen oxide concentration value, and the sub-PID controller is responsible for adjusting the execution equipment; In traditional denitrification treatment, it is difficult to strike a balance between efficiency and cost through a single control. However, the dual PID system can more accurately balance the two. Through the synergistic effect of the main and auxiliary PIDs, ammonia dosage is dynamically optimized, ensuring denitrification results while effectively reducing operating costs, maximizing resource utilization, improving overall system stability, and ensuring long-term efficient operation. S160. Establish a big data model library to store the data and adjustment methods of daily operation into the memory library, including inlet flue gas temperature, flow, initial NOx concentration and distribution, ammonia injection amount, intelligent valve adjustment status, outlet NOx concentration, outlet flue gas temperature, oxygen content, etc., and enhance the scale of the model to make subsequent operation more optimized and smarter, thus having self-learning function.

[0021] Example 2: Figure 2 The module structure diagram of the intelligent denitrification system according to the embodiment of the present application is schematically shown. Figure 2 As shown in the figure, the intelligent denitrification system includes the following modules: A distributed control system construction module is used to construct a distributed control system based on Modbus communication, open a Modbus communication interface, and collect data through an acquisition device. The collected data is the first collected data, and a data link is established for the first collected data. A denitrification artificial intelligence model creation module is used to create a denitrification artificial intelligence model. The denitrification artificial intelligence model has an input end and an output end. The input end is used to input raw data, and the output end is used to output denitrification effect prediction data processed by the model. A strategy model creation module is used to create a strategy model, which contains an outlet nitrogen oxide concentration prediction model and works in conjunction with the denitrification artificial intelligence model; An execution module is used to generate an execution signal and send the execution signal to an execution device, and the ammonia dosage is adjusted by the execution device, and a PID controller is used for regulation during the adjustment process; Closed-loop control module, used to establish a closed-loop control system and set up a main PID controller and a sub-PID controller. The main control PID controller is responsible for calculating the inlet NOx concentration value, the set value and the outlet NOx concentration value, and the sub-PID controller is responsible for executing the equipment adjustment. The big data model library establishment module is used to establish a big data model library, storing the normal operation data and adjustment methods into the memory library, including inlet flue gas temperature, flow, NOx initial concentration and distribution, ammonia injection amount, intelligent valve adjustment status, outlet NOx concentration, outlet flue gas temperature, oxygen content, and enhancing the scale of the model to make subsequent operation more optimized and smarter, thus having self-learning function.

[0022] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0023] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. The treatment method of the intelligent denitrification system is characterized by: The method includes: Constructing a distributed control system based on Modbus communication, opening a Modbus communication interface, and collecting data through an acquisition device, wherein the collected data is first collected data, and constructing a data link for the first collected data; Creating a denitrification artificial intelligence model, the denitrification artificial intelligence model having an input end and an output end, the input end being used to input raw data, and the output end being used to output denitrification effect prediction data processed by the model; Create a strategy model that includes an outlet NOx concentration prediction model and works in conjunction with the denitrification artificial intelligence model; Generate an execution signal and send it to the execution device, which adjusts the amount of ammonia solution added, using a PID controller for regulation during the adjustment process; Establish a closed-loop control system and set up a main PID controller and a sub-PID controller. The main PID controller is responsible for calculating the inlet NOx concentration value, the set value, and the outlet NOx concentration value, while the sub-PID controller is responsible for executing the equipment adjustment. A big data model library is established to store routine operation data and adjustment methods into the memory library, including inlet flue gas temperature, flow, initial NOx concentration and distribution, ammonia injection amount, intelligent valve adjustment status, outlet NOx concentration, outlet flue gas temperature, oxygen content, and enhance the scale of the model to make subsequent operations more optimized and smarter, thus having self-learning capabilities.

2. The method according to claim 1, characterized in that Data link construction includes: Use Modbus to read data from the distributed control system and then develop the Modbus interface; Conduct multiple rounds of stress testing on the interface software, simulating high-concurrency scenarios to verify its load-bearing capacity; The interface software is called by the first device, and data protocol conversion is performed after the data is read, and the called data is pre-processed, and then the pre-processed data is stored in the first device, where the first device is an RK3199 single-board computer; Create data storage and query structures and link them with system clients. System clients query real-time data and historical trends by selecting interface programs and database storage directories.

3. The method according to claim 2, characterized in that After the data protocol conversion, preprocessing operations are required, which include data noise reduction processing and 0-2 interpolation processing.

4. The method according to claim 1, wherein Creating a denitrification AI model further includes: A model input layer is set to receive raw data of the denitrification process, where the raw data is first collected data stored in the distributed control system and collected by the collection device; Set up the model's hidden layer and receive data from the model's input layer. The model's hidden layer uses a multi-layer neural network structure to extract features from the data layer by layer. The number of layers and nodes in the multi-layer neural network structure needs to be dynamically adjusted according to the actual data characteristics. Set up the model training and validation layers, create training sets and validation sets, perform feature normalization, then train the model using the training set and validate the model using the validation set.

5. The method according to claim 4, characterized in that In the training and validation layers, the training and validation processes are placed on the same level, and the bias is adjusted during the training and validation processes.

6. The method according to claim 4, characterized in that The outlet nitrogen oxide concentration prediction model has an input end, which is used to input the first collected data and the features extracted by the model hidden layer, and an output end, which is used to output the predicted outlet nitrogen oxide concentration. The nitrogen oxide controlled variable is generated by comparing and fitting the concentration with the actual collected outlet nitrogen oxide concentration.

7. The method according to claim 1, characterized in that Generate an execution signal and send the execution signal to the execution device, and link with the execution device through the execution signal, where the execution device is a pneumatic control valve or a combination of a pneumatic control valve and an electric control valve.

8. Intelligent denitrification system, characterized by: The system includes: A distributed control system construction module is used to construct a distributed control system based on Modbus communication, open a Modbus communication interface, and collect data through an acquisition device. The collected data is the first collected data, and a data link is established for the first collected data. A denitrification artificial intelligence model creation module is used to create a denitrification artificial intelligence model. The denitrification artificial intelligence model has an input end and an output end. The input end is used to input raw data, and the output end is used to output denitrification effect prediction data processed by the model. A strategy model creation module is used to create a strategy model, which contains an outlet nitrogen oxide concentration prediction model and works in conjunction with the denitrification artificial intelligence model; An execution module is used to generate an execution signal and send the execution signal to an execution device, and the ammonia dosage is adjusted by the execution device, and a PID controller is used for regulation during the adjustment process; Closed-loop control module, used to establish a closed-loop control system and set up a main PID controller and a sub-PID controller. The main control PID controller is responsible for calculating the inlet NOx concentration value, the set value and the outlet NOx concentration value, and the sub-PID controller is responsible for executing the equipment adjustment. The big data model library establishment module is used to establish a big data model library, storing the normal operation data and adjustment methods into the memory library, including inlet flue gas temperature, flow, NOx initial concentration and distribution, ammonia injection amount, intelligent valve adjustment status, outlet NOx concentration, outlet flue gas temperature, oxygen content, and enhancing the scale of the model to make subsequent operation more optimized and smarter, thus having self-learning function.

9. The intelligent denitrification system according to claim 8, characterized in that: A method for operating the intelligent denitrification system according to any one of claims 1 to 7.

Images