Hydrolyzer ammonia production and denitration whole-process cooperative control method based on NOx prediction

Abstract

The invention discloses a hydrolyzer ammonia production and denitration whole-process cooperative control method based on NOx prediction, and the method comprises the steps: collecting the real-time state parameters of the operation of a boiler, inputting the real-time state parameters into an LSTM neural network prediction model, and obtaining an SCR inlet NOx concentration prediction value which is earlier than the actual measurement; on the basis of the SCR inlet NOx concentration predicted value and in combination with the load change parameters of the boiler, an advanced ammonia demand signal is obtained through calculation; and synchronously generating an ammonia spraying optimization instruction for controlling an SCR denitration system and an ammonia production optimization instruction for controlling a hydrolyzer ammonia production system by using the advanced ammonia demand signal. The advanced ammonia demand signal capable of guiding the hydrolyzer ammonia production system and the SCR denitration system to carry out advanced actions at the same time is generated by calculating the NOx concentration predicted value at the SCR inlet, supply and demand cooperation and advanced response of the ammonia production and denitration links are achieved, and the problems of ammonia spraying control lag and ammonia supply pressure fluctuation caused by measurement delay and system isolation are effectively solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of flue gas purification and automatic control of thermal power plants, and particularly relates to a kind of based on NO x The present application relates to the field of flue gas purification and automatic control of thermal power plants, and particularly relates to a kind of based on NO x Predicted hydrolysis ammonia and denitration whole process collaborative control method. BACKGROUND

[0002] With the continuous growth of wind power, photovoltaic and other new energy installed capacity, thermal power units are gradually changing from the main power supply to auxiliary service type power supply. This change puts forward higher requirements for the operation of thermal power units: on the one hand, the unit needs to meet the demand of deep peak regulation and flexible operation; on the other hand, it still needs to cope with ultra-low emission and the increasingly stringent environmental protection index challenge of the local area, among which the treatment of nitrogen oxides (NO x ) is particularly important.

[0003] Through the research and analysis of the operation of the denitration system of coal-fired power plants, it is found that there are several outstanding problems in the existing technology. The denitration device of coal-fired power plants in China generally has the problems of excessive ammonia injection, low denitration efficiency, and instantaneous environmental protection index exceeding the standard. Especially during the substantial adjustment of the unit load, the urea hydrolysis ammonia system occasionally accompanies abnormal operation, resulting in unstable ammonia supply. At present, the reducing agent of the denitration system has gradually changed from liquid ammonia to urea ammonia with higher safety. However, the research on the whole process collaborative control of the hydrolysis ammonia and denitration system under complex working conditions is still insufficient in China, which easily leads to large pressure fluctuations of the product gas at the outlet of the hydrolysis reactor, delayed ammonia supply, and further affects the stable operation of the denitration system.

[0004] The existing ammonia injection control system mostly adopts a simple feedforward combined with PID control strategy, which calculates the feedforward ammonia demand through the parameters such as the NO x concentration at the inlet of SCR, flue gas volume and boiler load. However, due to the long response delay time of the NO x concentration measurement using the extraction sampling analysis method, it cannot timely and accurately adjust the ammonia injection amount when the unit load is rapidly increased or decreased, resulting in severe fluctuations of the NO x concentration at the outlet of the denitration.

[0005] The existing ammonia injection control strategy cannot adapt to the new demand of deep peak regulation and flexible operation of thermal power units. Therefore, developing an advanced technology that can realize the whole process collaborative control of the hydrolysis ammonia and SCR denitration based on the NO x prediction is of great significance for balancing the safety operation of the unit and the environmental protection emission demand, reducing ammonia escape and the NO xIt is particularly important and urgent to provide a high-efficiency, stable and economical NOx treatment solution for the thermal power industry by preventing the concentration fluctuation and the "false liquid level" safety risk of the hydrolyzer. SUMMARY

[0006] The purpose of the embodiment of the present application is to provide a NOx x The predicted hydrolyzer ammonia production and denitration whole-process collaborative control method acquires the predicted value of the NOx concentration at the SCR inlet in advance through the LSTM neural network prediction model, generates an advanced ammonia demand signal which can simultaneously guide the advanced action of the hydrolyzer ammonia production system and the SCR denitration system, realizes the supply-demand collaboration and advanced response of the ammonia production and denitration links, and effectively solves the problems of ammonia injection control lag and ammonia supply pressure fluctuation caused by measurement delay and system isolation. x

[0007] To solve the above technical problems, the first aspect of the embodiment of the present application provides a NOx x predicted hydrolyzer ammonia production and denitration whole-process collaborative control method, which comprises the following steps: Collecting real-time state parameters of the boiler operation, inputting the real-time state parameters into an LSTM neural network prediction model, and acquiring a predicted value of the NOx concentration at the SCR inlet in advance of actual measurement; x Based on the predicted value of the NOx concentration at the SCR inlet, in combination with the load variation parameters of the boiler, an advanced ammonia demand signal is calculated, which is used to simultaneously guide the advanced action of the hydrolyzer ammonia production system and the SCR denitration system; x Using the advanced ammonia demand signal, an ammonia injection optimization instruction for controlling the SCR denitration system and an ammonia production optimization instruction for controlling the hydrolyzer ammonia production system are synchronously generated.

[0008] Further, the using of the advanced ammonia demand signal to synchronously generate the ammonia injection optimization instruction for controlling the SCR denitration system and the ammonia production optimization instruction for controlling the hydrolyzer ammonia production system comprises: Real-time collection of the boiler load change rate, the load target value and the predicted value of the NOx concentration at the SCR inlet, and identification of the operating mode of the current operation state of the unit; x According to the identified operating mode, the control priority is dynamically determined, and the multi-target coordinator is internally provided with an optimization strategy corresponding to each operating mode; Based on the dynamically determined control priority and the advanced ammonia demand signal, a first control signal for the SCR denitration system and a second control signal for the hydrolyzer ammonia production system are respectively generated; ​​​​The ammonia injection optimization instruction is generated based on the first control signal, and the ammonia production optimization instruction is generated based on the second control signal, so as to realize the collaborative control of the SCR denitration system and the hydrolyzer ammonia production system.

[0009] Further, the control priority is dynamically determined according to the identified working condition mode, including: Based on the working condition mode recognition result, a key performance indicator threshold corresponding to the current unit operating state is obtained; According to the key performance indicator threshold, an optimization strategy matched with the current working condition mode is selected through an expert rule library built in the multi-objective coordinator; According to the selected optimization strategy, a dynamic weight parameter for coordinated control is determined, and the dynamic weight parameter includes an SCR side control priority coefficient and a hydrolyzer side control priority coefficient.

[0010] Further, the key performance indicator threshold corresponding to the current unit operating state is obtained based on the working condition mode recognition result, including: According to the working condition mode recognition result, a basic threshold of the key performance indicator is read from a preset working condition-threshold mapping relationship, and the basic threshold includes a basic ammonia escape concentration upper limit value, a basic SCR outlet NO x Concentration upper limit value and basic hydrolyzer liquid level safety range boundary value; Real-time monitoring of unit operating state parameters, including boiler load rate, boiler load change rate, economizer outlet oxygen content, total air volume to total coal volume ratio; Based on the real-time monitored unit operating state parameters, a corresponding threshold correction coefficient is queried through a preset threshold correction coefficient table, and the threshold correction coefficient table stores correction coefficient values corresponding to different unit operating state parameter intervals; The basic threshold is multiplied by the corresponding threshold correction coefficient to obtain the key performance indicator threshold corresponding to the current unit operating state.

[0011] Further, the first control signal for the SCR denitration system and the second control signal for the hydrolyzer ammonia production system are respectively generated based on the dynamically determined control priority and the lead ammonia demand signal, including: Based on the control priority, an SCR side amplitude coefficient and a hydrolyzer side amplitude coefficient are obtained; Based on the lead ammonia demand signal, an SCR side reference control signal is calculated in combination with the SCR side amplitude coefficient, and a hydrolyzer side reference control signal is calculated in combination with the hydrolyzer side amplitude coefficient; According to the control priority, a preset response rate configuration table is queried to obtain an SCR side maximum change rate limit value and a hydrolyzer side maximum change rate limit value; According to the SCR side maximum change rate limit value, a change rate limiting process is performed on the SCR side initial control amount, and a processed result is taken as the first control signal; According to the hydrolyzer side maximum change rate limit value, a change rate limiting process is performed on the hydrolyzer side initial control amount, and a processed result is taken as the second control signal.

[0012] Further, the SCR inlet NO x The concentration prediction value is combined with a load variation parameter of the boiler to calculate a lead ammonia demand signal, including: A first key parameter reflecting a dynamic characteristic of the hydrolyzer ammonia production system is obtained, and the first key parameter includes a hydrolyzer liquid level change rate, a real-time measured value of a hydrolyzer outlet pressure and a urea solution supply flow; A second key parameter reflecting an operation state of the SCR denitration system is obtained, and the second key parameter includes an SCR inlet flue gas temperature change rate and a real-time measured value of a pressure difference between the SCR reactor inlet and outlet; A boiler total air volume, a boiler total coal volume and a coal economizer outlet oxygen volume are obtained, and a dynamic flue gas flow parameter is calculated; The first key parameter of the hydrolyzer ammonia production system and the second key parameter of the SCR denitration system are input into a collaborative calculation model to obtain a system dynamic coordination factor; Based on the load variation parameter of the boiler and the system dynamic coordination factor, the SCR inlet NO x concentration prediction value is collaboratively corrected; Based on the collaboratively corrected SCR inlet NO x concentration prediction value, the dynamic flue gas flow parameter is fused to generate the lead ammonia demand signal.

[0013] Further, the first key parameter of the hydrolyzer ammonia production system and the second key parameter of the SCR denitration system are input into a collaborative calculation model to obtain a system dynamic coordination factor, including: A hydrolyzer operation stability index is calculated based on the first key parameter, an SCR operation stability index is calculated based on the second key parameter, a ratio of the hydrolyzer operation stability index to the SCR operation stability index is calculated, and a system stability ratio is obtained; The coordination factor reference value is calculated by a segmented linear function based on the system stability ratio and the current load rate of the unit, the coordination factor reference value is multiplied by a dynamic adjustment coefficient to obtain a preliminary system dynamic coordination factor, wherein the dynamic adjustment coefficient is calculated according to the boiler load change rate and the oxygen content change rate at the outlet of the coal economizer; The preliminary system dynamic coordination factor is subjected to nonlinear amplitude limiting processing to ensure that the value is within a preset effective range, and the processed result is output as the final system dynamic coordination factor.

[0014] Further, based on the load variation parameter of the boiler and the system dynamic coordination factor, the SCR inlet NO x Concentration prediction value is cooperatively corrected, including: A load variation influence degree is calculated based on the boiler load change rate, a load target tracking coefficient is calculated based on the deviation between the boiler load target value and the current load value, and the two are weighted and fused to obtain a load dynamic correction coefficient; The system dynamic coordination factor is used as a correction weight to weight and adjust the load dynamic correction coefficient to obtain a cooperative correction coefficient; The SCR inlet NO x Concentration prediction value is multiplied by the cooperative correction coefficient to obtain a cooperatively corrected NO x Concentration value, and subjected to saturation amplitude limiting processing to obtain the corrected SCR inlet NO x Concentration prediction value.

[0015] Further, the real-time state parameters of the boiler operation are collected, the real-time state parameters are input into an LSTM neural network prediction model to obtain the SCR inlet NO x Concentration prediction value in advance of actual measurement, including: Real-time state parameters in the operation process of the boiler are collected, and the real-time state parameters include total air volume, total coal volume, main steam flow, flue gas temperature at the outlet of the furnace, secondary air volume of each layer, overfire air volume, and oxygen content at the outlet of the coal economizer; The real-time state parameters are subjected to data preprocessing to form a time series data set; Based on the LSTM neural network prediction model, the NO x Concentration estimation value is calculated in combination with the time series data set; The NOx concentration estimation value is weighted and fused with the current SCR inlet NO x Concentration measured value to obtain the SCR inlet NO x Concentration prediction value.

[0016] Further, based on the LSTM neural network prediction model, the NOx The concentration estimation value comprises: The time series data set is organized into a model input sequence according to a fixed time step, and the model input sequence is input into the LSTM neural network prediction model; The model input sequence is processed by the encoder of the LSTM neural network prediction model in a time step, and each LSTM unit of the encoder calculates the hidden state of the current time according to the input of the current time step and the hidden state of the previous time; The hidden state of the last time step of the encoder is input into the decoder of the LSTM neural network prediction model as a context vector, and the NO x The concentration estimation value.

[0017] Correspondingly, a second aspect of the embodiment of the application provides an electronic device, comprising: at least one processor; and a memory connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above-mentioned NO x The predicted hydrolysis reactor ammonia production and denitration whole process collaborative control method.

[0018] Correspondingly, a third aspect of the embodiment of the application provides a computer readable storage medium, which stores computer instructions, and the instructions are executed by a processor to implement the above-mentioned NO x The predicted hydrolysis reactor ammonia production and denitration whole process collaborative control method.

[0019] The above technical solutions of the embodiment of the application have the following beneficial technical effects: 1. By establishing an LSTM neural network prediction model based on an LSTM neural network, the response delay problem existing in traditional extractive CEMS measurement devices is effectively overcome, and the change trend of the SCR inlet NO x The concentration can be obtained in advance and accurately, which provides a key leading decision basis for subsequent ammonia demand calculation and whole process collaborative control, thereby significantly improving the timeliness and accuracy of ammonia injection control, and enabling the denitration system to maintain stable outlet NO x Concentration under the condition of rapid load change of the unit, avoiding the risk of environmental protection index fluctuation and exceeding standard due to control lag; 2. A full-process collaborative control architecture covering ammonia production by hydrolysis and SCR denitration is designed, the control priority is dynamically determined by a multi-objective coordinator, and the single lead ammonia demand signal is converted into control instructions for guiding the collaborative action of the two systems by a collaborative control signal generator, thereby fundamentally solving the "control island" problem of independent ammonia production and denitration in traditional control, achieving precise matching and linked response of the ammonia "production, supply and use" links, and greatly relieving the imbalance between supply and demand and the outlet pressure fluctuation of the hydrolysis device caused by sudden load changes; 3. By constructing a full-process control strategy based on dynamic weight parameters and collaborative correction mechanism, adaptive collaborative optimization of the hydrolysis ammonia production system and the SCR denitration system under different operating conditions is achieved; the amplitude and response rate of the control signal can be dynamically adjusted according to real-time load changes, key performance indicator thresholds and system operating state, so that the SCR system with fast response and the hydrolysis device with large inertia can reach the best cooperation state, thereby effectively reducing ammonia consumption, prolonging the service life of the catalyst, and significantly improving the operating economy and stability of the entire denitration system while ensuring denitration efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The present application provides a NO x The flow chart of the hydrolysis ammonia production and denitration full-process collaborative control method based on NO DETAILED DESCRIPTION

[0021] In order to make the objectives, technical solutions and advantages of the present application clearer, further detailed description will be made below in combination with specific embodiments and with reference to the drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present application. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present application.

[0022] Please refer to Figure 1 To solve the above technical problems, the first aspect of the present application provides a NO x The hydrolysis ammonia production and denitration full-process collaborative control method based on NO Step S100, collecting real-time state parameters of the boiler operation, inputting the real-time state parameters into the LSTM neural network prediction model, and obtaining the predicted value of the SCR inlet NO x concentration in advance of actual measurement.

[0023] In the deep peak regulation operation process of large coal-fired units, the boiler combustion state presents a rapid dynamic change characteristic. In order to realize the prediction of the SCR inlet NO xThe advanced prediction of the concentration requires the collection of second-level real-time parameters reflecting the combustion state, including macro operating parameters such as total air volume, total coal volume, main steam flow, and refined air distribution parameters such as secondary air volume of each layer and overfire air volume, as well as key state indicators such as flue gas temperature at the furnace outlet and oxygen content at the economizer outlet. These parameters collectively constitute a multi-dimensional data set representing the boiler combustion state. By analyzing the time sequence variation of the data set, the NOx x precursor features can be effectively captured.

[0024] The collected real-time state parameters are input into an LSTM neural network prediction model trained with a large amount of historical operating data. The model can learn the long-term dependence features of the boiler operating parameters and NOx x concentration through its unique gating mechanism, and can generate a complex nonlinear mapping relationship between the parameters and remember the long-term dependence features of the parameter changes. In actual applications, the input sequence length of the model is usually set to cover a complete load variation period to ensure that the complete dynamic process can be captured. Through forward calculation of the model, the NOx x concentration prediction value can be output 60-90 seconds in advance of the actual CEMS measurement value, which provides a sufficient response window for the subsequent control link.

[0025] In step S200, the predicted NOx x concentration at the SCR inlet is obtained based on the predicted NOx x concentration at the SCR inlet, the boiler load variation parameters, and the ammonia demand calculation model. The advanced ammonia demand signal is used to guide the simultaneous action of the hydrolyzer ammonia production system and the SCR denitration system.

[0026] Based on the predicted NOx x concentration at the SCR inlet, the boiler load variation rate, the load target value, and other parameters representing the unit operating state, the advanced ammonia demand signal is obtained through the established ammonia demand calculation model. This calculation process not only considers the NOx x concentration prediction value at the current time, but also introduces a dynamic compensation mechanism during the load variation process to ensure that the ammonia demand is increased in advance when the unit load is increased, and the ammonia supply is reduced in time when the unit load is decreased, avoiding the imbalance between supply and demand.

[0027] The obtained advanced ammonia demand signal is sent to the hydrolyzer ammonia production system and the SCR denitration system as a unified command signal. For the hydrolyzer system, the signal guides the adjustment of the urea solution feed amount and the steam supply amount in advance to prepare for the subsequent ammonia demand changes; for the SCR denitration system, the signal serves as the basis for feedforward control, guiding the ammonia injection grid to adjust the ammonia injection preparation state in advance. This dual-system synchronous guidance mechanism ensures the dynamic balance between ammonia supply and consumption.

[0028] Step S300, using the lead ammonia demand signal, synchronously generates an ammonia injection optimization instruction for controlling the SCR denitration system and an ammonia production optimization instruction for controlling the hydrolyzer ammonia production system.

[0029] After receiving the lead ammonia demand signal, the multi-objective coordinator dynamically determines the control priority according to the current operating condition. In the rapid load variation stage, the response speed of the SCR denitration system is prioritized; in the stable load stage, the operating economy of the hydrolyzer ammonia production system is focused on. This priority allocation is realized by adjusting the amplitude coefficient and the change rate limit of the control signal, ensuring that the two systems are in their respective safe operating ranges and achieve optimal cooperation.

[0030] Based on the determined control strategy, the cooperative control signal generator generates an ammonia injection optimization instruction for the SCR denitration system and an ammonia production optimization instruction for the hydrolyzer ammonia production system. The ammonia injection optimization instruction can be adjusted by adjusting the two side main ammonia injection valves to achieve rapid control; the ammonia production optimization instruction determines the optimal opening combination of the urea solution feed valve, the steam feed valve and the product gas outlet valve through material balance and energy balance calculation. The control instructions of the two systems maintain strict synchronization and coordination, realizing the whole process closed-loop optimization from ammonia production to consumption.

[0031] By establishing a complete technical chain from NO x x concentration prediction to ammonia demand calculation, and then to whole process cooperative control, the precise cooperation and dynamic optimization of the hydrolyzer ammonia production system and the SCR denitration system in the deep peak regulation condition are realized, effectively solving the response lag and supply-demand imbalance problems existing in the traditional control method, significantly improving the control quality and operating economy of the denitration system, and ensuring the safe and stable operation of the system.

[0032] Specifically, in step S100, the real-time state parameters of the boiler operation are collected, the real-time state parameters are input into the LSTM neural network prediction model, and the SCR inlet NO x x concentration prediction value is obtained, which includes: Step S110, collecting real-time state parameters during the operation of the boiler, the real-time state parameters including total air volume, total coal volume, main steam flow, furnace outlet flue gas temperature, each layer secondary air volume, overfire air volume and economizer outlet oxygen concentration.

[0033] In the actual operation of large coal-fired units, the monitoring of the boiler combustion state needs to cover the whole process from fuel input to flue gas emission. Taking a 1050 MW ultra-supercritical unit as an example, the data acquisition system needs to obtain the total air volume and the total coal volume in real time, which are two core parameters that directly determine the basic operating conditions of the furnace combustion. At the same time, the main steam flow, as a direct reflection of the unit load, its trend reflects the energy output state of the boiler. The monitoring of the flue gas temperature at the furnace outlet is particularly important, which can directly reflect the temperature distribution of the combustion center, and the temperature level is related to the generation path of NO x There is a clear correlation between the generation rate.

[0034] In addition to the above macro parameters, the distribution data of the secondary air volume and the overfire air volume of each layer need to be collected in detail, which reflects the way of combustion organization and the effect of staged combustion. In actual operation, the opening regulation of the secondary air damper of each layer will significantly affect the oxygen concentration distribution of the combustion zone, thereby changing the generation path of NO x . The oxygen concentration at the economizer outlet, as an important indicator to measure the excess air coefficient, its value change is directly related to the generation amount of fuel-type NO x . These parameters together constitute a complete combustion state monitoring system, providing a comprehensive data basis for subsequent NO x concentration prediction.

[0035] Step S120, data preprocessing is performed on the real-time state parameters to form a time series data set.

[0036] The collected original operating parameters need to go through a strict data preprocessing process to ensure the data quality input into the prediction model. First, abnormal value detection and elimination are performed on each parameter, for example, when the value of a certain measuring point exceeds the range of the sensor or is obviously contradictory to the associated parameters, the data at that time will be marked as abnormal and excluded. At the same time, since there may be differences in the sampling frequency of different parameters, data alignment processing is needed to unify all parameters to the same time stamp to form a complete time series data set.

[0037] On the basis of data alignment, normalization processing is also needed to eliminate the influence of different parameter dimensions. In specific implementation, the maximum and minimum value normalization method is adopted to map the values of each parameter to the interval of zero to one. This processing not only improves the convergence speed of model training, but also avoids the problem that some large value parameters dominate the model calculation. The standardized time series data set formed after preprocessing not only retains the change rule of the original parameters, but also meets the requirements of the LSTM neural network model for input data format, laying a solid foundation for subsequent accurate prediction.

[0038] Step S130, based on the LSTM neural network prediction model, the time series data set is combined to calculate the NOx Concentration estimate.

[0039] The LSTM neural network prediction model is built upon training with a large amount of historical operational data, and its structure has been specifically optimized to adapt to the characteristics of the boiler combustion process. In practical applications, the model's input layer is designed as a multivariate time series interface, capable of simultaneously receiving time series data of seven key parameters, including total air volume and total coal volume. The hidden layers employ a two-layer LSTM structure, with each layer containing fifty neurons. This deep structure effectively captures the complex nonlinear relationships and long-term temporal dependencies between parameters.

[0040] The model training process employs a sliding time window mechanism, with a window length of 120 sampling points, corresponding to two minutes of real-time data. In each prediction period, the model receives the latest time-series data set, filters and memorizes key information through the gating mechanism of the LSTM unit, and finally outputs the NO value for the next 90 seconds. x Concentration prediction. The prediction time span is carefully designed to take into account both the transmission delay of flue gas from the furnace to the SCR reaction zone and to allow sufficient response time for subsequent control steps, ensuring that the prediction results have practical guiding value.

[0041] Step S140, NO x Concentration forecast and current SCR inlet NO x The measured concentration values ​​were weighted and fused to obtain the SCR inlet NO. x Predicted concentration values.

[0042] To improve the reliability and adaptability of the predictions, a weighted fusion mechanism between predicted and measured values ​​was designed. In actual operation, the NO output of the LSTM model... x The concentration prediction is compared and fused in real time with the measured value from the CEMS analyzer installed at the SCR inlet. The fusion weight is determined based on the confidence level of the model prediction results and the reliability assessment of the measured data. When the confidence level of the model prediction is high, the weight of the predicted value is appropriately increased; when the CEMS measurement data is stable and reliable, the weight of the measured value is increased.

[0043] This weighted fusion mechanism effectively improves the robustness of the system, maintaining relatively accurate NO even in the event of temporary malfunctions or measurement anomalies in individual sensors. x Concentration output. In practice, a dynamic weight adjustment algorithm is set up to automatically adjust the fusion ratio based on historical statistics of prediction errors and current data quality. The SCR input NO after fusion processing... x The concentration predictions retain the advantage of the model's predictive power while incorporating the real-time accuracy of measured data, providing a more reliable data foundation for subsequent ammonia demand calculations.

[0044] By establishing a complete boiler operating parameter acquisition system, combining a specially optimized LSTM neural network prediction model and an intelligent data fusion mechanism, the NOx concentration at the SCR inlet is predicted x The accurate and advanced prediction of the concentration provides key technical support for subsequent ammonia demand calculation and full-process collaborative control, effectively solves the response lag problem existing in traditional measurement methods, and significantly improves the forward-looking control capability of the denitration system.

[0045] Further, in step S130, the LSTM neural network prediction model is used to calculate the NOx concentration prediction value based on the time series data set, including: x In step S131, the time series data set is organized into a model input sequence according to a fixed time step, and the model input sequence is input into the LSTM neural network prediction model.

[0046] In actual operation of the unit, the time series data set needs to be organized into a model input sequence according to a fixed time step, which is usually set to 1 second to match the fast dynamic characteristics of the boiler combustion process. The length of the model input sequence needs to cover a complete load response period, usually set to 120 time steps, i.e. 2 minutes of real-time data. Each time step contains the normalized values of 7 key parameters such as total air volume, total coal volume and main steam flow, forming a multi-dimensional input vector. In actual application, the input sequence is updated in real time through a sliding window mechanism to ensure that the model always predicts based on the latest operating state. When the organized model input sequence is input into the LSTM neural network prediction model, the model first extracts features from the input data, identifies the spatio-temporal correlation between parameters through its internal gating mechanism, and provides a structured data basis for subsequent sequence encoding.

[0047] In step S132, the model input sequence is processed by the encoder of the LSTM neural network prediction model step by step, and each LSTM unit of the encoder calculates the hidden state at the current time step based on the input at the current time step and the hidden state at the previous time step.

[0048] The encoder of the LSTM neural network prediction model consists of two layers of LSTM units, each containing 64 neurons, for processing the input sequence step by step. Each LSTM unit calculates the hidden state and cell state at the current time step based on the input vector at the current time step and the hidden state at the previous time step through the coordinated action of the input gate, the forget gate and the output gate. In actual operation, the forget gate determines the proportion of historical information to be retained according to the current input and the previous state, the input gate controls the degree of update of new information, and the output gate adjusts the output content of the hidden state. This mechanism enables the model to effectively capture the long-term dependence of parameter changes in the combustion process, such as the adjustment of the air-coal ratio on the NOx concentration, and the influence of the main steam flow on the NOx concentration. x ​The generated delay effect. The encoder encodes the timing features of the entire input sequence into a series of hidden states through iterative processing by time step, where the hidden state of the last time step contains the condensed information of the entire sequence.

[0049] Step S133, the hidden state of the last time step of the encoder is input as a context vector to the decoder of the LSTM neural network prediction model, and the NO x concentration prediction value.

[0050] The hidden state of the last time step of the encoder is input as a context vector to the decoder, which carries all the timing feature information of the input sequence. The decoder also adopts a two-layer LSTM structure, which generates the prediction sequence step by step through its internal LSTM units after receiving the context vector. In practical applications, the first LSTM unit of the decoder takes the context vector as the initial input, and the subsequent time steps receive the output of the previous time step in turn. The decoder output of each time step is linearly transformed by a fully connected layer to map high-dimensional features to specific NO x concentration prediction value. The fully connected layer uses a linear activation function, and the output dimension is set to the number of prediction time points, usually set to the prediction value every 10 seconds within the next 90 seconds. This encoder-decoder structure can effectively establish the complex mapping relationship between the boiler operating parameters and the future NO x concentration, achieving accurate multi-step prediction.

[0051] Through the LSTM neural network prediction model based on the encoder-decoder structure, accurate multi-step prediction of the NO x concentration at the SCR inlet is realized, and the model can effectively learn the complex nonlinear relationship between the boiler operating parameters and the NO x concentration generation, and capture the long-term timing features of parameter changes, providing reliable advance prediction information for subsequent ammonia demand calculation and whole-process collaborative control, significantly improving the feedforward control capability of the denitration system.

[0052] Specifically, based on the predicted value of the NO x concentration at the SCR inlet in step S200, the advance ammonia demand signal is calculated by combining the load variation parameters of the boiler, including: Step S210, obtain the first key parameter reflecting the dynamic characteristics of the hydrolysis reactor ammonia production system, the first key parameter includes the real-time measurement value of the hydrolysis reactor liquid level change rate, the hydrolysis reactor outlet pressure and the urea solution supply flow.

[0053] The dynamic characteristics monitoring of the ammonia production system of the hydrolyzer needs to focus on three key operating parameters. The change rate of the hydrolyzer liquid level reflects the balance of urea solution consumption and supply. When the liquid level change rate exceeds the set threshold, it indicates that the system may have uneven feeding or abnormal reaction rate. Real-time measurement of the hydrolyzer outlet pressure directly represents the stability of the ammonia supply system. Large pressure fluctuations may affect the stable operation of the subsequent SCR denitration system. The urea solution supply flow is the basic input parameter of the ammonia production system. Its value directly determines the theoretical yield of ammonia gas and needs to be kept within a reasonable range matching the unit load. These three parameters together constitute the core index system of the ammonia production system operation state monitoring, providing important state feedback information for subsequent collaborative control.

[0054] In actual operation, the change rate of the hydrolyzer liquid level is monitored in real time by a differential pressure transmitter with a sampling frequency of one second to ensure that rapid liquid level fluctuations can be captured. The hydrolyzer outlet pressure is measured by a high-precision pressure transmitter with a measurement range of zero to 1.6 MPa and an accuracy level of 0.5, which can accurately reflect the pressure state of the product gas pipe network. The urea solution supply flow is detected by an electromagnetic flowmeter installed on the hydrolyzer feed pipe to monitor the supply of 50% concentration urea solution in real time. Real-time data of these parameters are transmitted to the control system through the field bus, providing accurate data support for the state assessment of the ammonia production system.

[0055] Step S220, obtaining the second key parameter reflecting the operating state of the SCR denitration system, the second key parameter including the change rate of the SCR inlet flue gas temperature and the measured value of the pressure difference between the inlet and outlet of the SCR reactor.

[0056] The operating state monitoring of the SCR denitration system focuses on two representative parameters. The change rate of the SCR inlet flue gas temperature reflects the influence of boiler load changes on the flue gas system. Rapid temperature changes may affect the reaction efficiency and life of the catalyst. The measured value of the pressure difference between the inlet and outlet of the SCR reactor directly reflects the plugging condition of the catalyst layer. Abnormal increase in pressure difference usually means that catalyst dusting or plugging is aggravated, which needs to be adjusted or maintained in time. These two parameters can effectively represent the operating health status of the SCR system, providing an important basis for the optimization control of the denitration system.

[0057] In the implementation process, the SCR inlet flue gas temperature is measured by a K-type thermocouple, and the temperature measuring point is located in the flue between the economizer outlet and the SCR reactor inlet, to ensure that the measurement result can truly reflect the flue gas temperature entering the catalyst. The pressure difference between the inlet and outlet of the SCR reactor is monitored by a differential pressure transmitter, and the pressure points are set in the straight pipe sections of the reactor inlet and outlet flues respectively to avoid the influence of air flow disturbance on the measurement accuracy. All measurement signals are transmitted by four to twenty milliamperes standard signals, and the sampling period is set to one second, which is uploaded to the control unit in real time through the data acquisition system, to provide timely and accurate data support for the state analysis and coordinated control of the system.

[0058] In step S230, the total air volume of the boiler, the total coal volume of the boiler and the oxygen content at the outlet of the economizer are obtained, and the dynamic flue gas flow parameter is calculated.

[0059] The calculation of the dynamic flue gas flow parameter needs to be based on three basic operating parameters: the total air volume of the boiler, the total coal volume of the boiler and the oxygen content at the outlet of the economizer. The total air volume reflects the total amount of air sent into the furnace, which is obtained through the air volume measuring device at the outlet of the air supply fan; the total coal volume represents the fuel input of the boiler, which is obtained through the rotation speed of the coal feeder and the coal quantity measuring device; and the oxygen content at the outlet of the economizer represents the excess air coefficient after combustion, which is monitored in real time by the zirconia oxygen analyzer. These three parameters together determine the actual flue gas generation amount and are the key basic data for calculating the ammonia requirement.

[0060] In the actual calculation process, the dynamic flue gas flow parameter is calculated based on the mass balance principle. First, the theoretical flue gas volume is calculated according to the total coal volume and the coal quality analysis data, and then the excess air coefficient is corrected in combination with the measured values of the total air volume and the oxygen content at the outlet of the economizer, to finally obtain the real-time total wet flue gas volume. The influence of environmental temperature and atmospheric pressure needs to be considered in the calculation process, and necessary compensation correction is made to the measured values. The obtained dynamic flue gas flow parameter is in units of standard cubic meters per hour, and is updated every second to ensure the accuracy and real-time nature of the subsequent ammonia requirement calculation. This parameter not only reflects the current flue gas working condition, but also provides an important reference for predicting the flue gas volume change trend during load variation.

[0061] In step S240, the first key parameter of the hydrolysis ammonia production system and the second key parameter of the SCR denitration system are input into the collaborative calculation model to obtain the system dynamic coordination factor.

[0062] The collaborative computing model's input includes three primary key parameters of the hydrolyzer ammonia production system and two secondary key parameters of the SCR denitrification system. The model first standardizes each parameter to eliminate dimensional differences, and then assigns corresponding weight coefficients based on the parameter's impact on system stability. The hydrolyzer level change rate and outlet pressure have higher weights, reflecting the importance of the ammonia production system's safe operation; the SCR reactor pressure difference and flue gas temperature change rate have lower weights, reflecting the need for performance optimization of the denitrification system. The weight coefficients are set based on the analysis results of a large amount of historical operating data and are dynamically adjusted according to the unit's operating status in practical applications.

[0063] The model calculates the system's dynamic coordination factor using a weighted algorithm. This factor ranges from zero to one, with a higher value indicating better system coordination. When the hydrolyzer system is operating stably and the SCR system is in good condition, the coordination factor approaches one; when either system experiences an anomaly, the coordination factor decreases accordingly. A correction for the load change rate is also incorporated into the calculation process, appropriately reducing the expected value of the coordination factor during periods of rapid load fluctuation to accommodate the system's dynamic adjustment. The final system dynamic coordination factor is updated every five seconds, serving as a crucial basis for subsequent coordinated adjustments and ensuring the control system can make reasonable decisions based on the actual operating conditions.

[0064] Step S250: Based on the boiler load variation parameters and system dynamic coordination factor, adjust the SCR inlet NO... x Concentration predictions are collaboratively corrected.

[0065] The coordinated correction process mainly relies on boiler load variation parameters and system dynamic coordination factors to adjust the SCR inlet NO. x The concentration forecast values ​​are optimized and adjusted. Load variation parameters include the load change rate and the load target deviation. The load change rate reflects the current operating trend of the unit, while the load target deviation characterizes the gap with the expected operating state. When the load change rate is large, the NO concentration is appropriately increased. x The correction range for the concentration prediction value is adjusted to cope with rapid changes in combustion conditions; when the load target deviation is large, the prediction value is adjusted accordingly based on the direction of the deviation.

[0066] The system dynamic coordination factor plays a role in regulating the correction intensity during this process. A higher coordination factor indicates good coordination between the ammonia production and denitrification systems, thus reducing the impact on NO. x The correction for the concentration prediction was relatively mild; when the coordination factor was low, it indicated a mismatch between systems, requiring a stronger correction to prevent control failure due to system incoordination. (Corrected NO) x The concentration prediction values ​​have been limited to ensure that they are within a reasonable range, neither too conservative nor too aggressive, thus providing more accurate and reliable input data for subsequent ammonia demand calculations.

[0067] Step S260, based on the synergistically corrected SCR inlet NOx concentration prediction value and the dynamic flue gas flow parameter, the ammonia demand signal is generated. x The concentration prediction value is fused with the dynamic flue gas flow parameter to generate the ammonia demand signal.

[0068] The generation of the ammonia demand signal is based on the synergistically corrected SCR inlet NOx concentration prediction value and the dynamic flue gas flow parameter. x The fusion calculation of the concentration prediction value and the dynamic flue gas flow parameter. First, according to the corrected NOx concentration prediction value, the molar amount of ammonia required for theoretical denitration is calculated, and then combined with the dynamic flue gas flow parameter to convert the molar amount into the actual ammonia mass flow demand. x The concentration prediction value is fused with the dynamic flue gas flow parameter to generate the ammonia demand signal. x The concentration prediction value is fused with the dynamic flue gas flow parameter to generate the ammonia demand signal. x The concentration prediction value is fused with the dynamic flue gas flow parameter to generate the ammonia demand signal.

[0069] In specific implementation, the ammonia demand calculation also includes the consideration of safety margin. According to the unit operation state and environmental protection requirements, a safety margin of five to ten percent is added to the theoretical calculation value to ensure the denitration effect while avoiding excessive ammonia escape. The final generated ammonia demand signal is in units of kilograms per hour, updated every ten seconds, and output to the hydrolyzer ammonia production system and the SCR denitration system. This signal not only contains the current ammonia demand information, but also predicts the demand change in the future through trend analysis, providing clear guidance for the coordinated operation of the two systems.

[0070] Further, in step S240, the first key parameter of the hydrolyzer ammonia production system and the second key parameter of the SCR denitration system are input into the synergistic calculation model to obtain the system dynamic coordination factor, including: Step S241, based on the first key parameter, the hydrolyzer operation stability index is calculated, based on the second key parameter, the SCR operation stability index is calculated, and the ratio of the hydrolyzer operation stability index and the SCR operation stability index is calculated to obtain the system stability ratio.

[0071] The calculation of the hydrolyzer operation stability index is based on a weighted comprehensive evaluation of first key parameters, in which the hydrolyzer liquid level change rate has the highest weight (0.5) as it directly reflects the system material balance state; the hydrolyzer outlet pressure real-time measurement value has a weight of 0.3, which embodies the stability of the ammonia supply system; and the urea solution supply flow has a weight of 0.2, which represents the system input stability. Each parameter is first normalized to map the actual value to the 0-1 interval, and the liquid level change rate is processed using an inverse proportional function after taking the absolute value to ensure that the smaller the change rate, the higher the score. The calculation of the SCR operation stability index uses different weight distribution, in which the SCR inlet flue gas temperature change rate has a weight of 0.6 as it directly affects the catalyst reaction efficiency; and the SCR reactor inlet and outlet pressure difference real-time measurement value has a weight of 0.4, which reflects the catalyst layer permeability state. After the two stability indexes are calculated respectively, the system stability ratio is obtained by taking the ratio of the two, and the ratio greater than 1 indicates that the hydrolyzer system is relatively stable, and the ratio less than 1 indicates that the SCR system is more stable, which provides a quantitative basis for subsequent coordinated control.

[0072] In step S242, based on the system stability ratio and the current load rate of the unit, the coordinated factor reference value is calculated by a segmented linear function, and the preliminary system dynamic coordinated factor is obtained by multiplying the coordinated factor reference value by the dynamic adjustment coefficient, wherein the dynamic adjustment coefficient is calculated according to the boiler load change rate and the oxygen content change rate at the outlet of the coal economizer.

[0073] The calculation of the coordinated factor reference value uses a three-segment segmented linear function, which is processed according to the interval in which the system stability ratio is located. When the stability ratio is in the interval of 0.8-1.2, the reference value remains 0.8; when the ratio is less than 0.8, the reference value decreases linearly with the decrease of the ratio, and the minimum value is 0.3; when the ratio is higher than 1.2, the reference value increases linearly with the increase of the ratio, and the maximum value is 1.0. At the same time, the current load rate of the unit is introduced for correction, which remains unchanged in the 70%-100% load interval, is multiplied by a coefficient of 0.9 in the 50%-70% load interval, and is multiplied by a coefficient of 0.8 when the load is less than 50%. The dynamic adjustment coefficient is determined by the boiler load change rate and the oxygen content change rate at the outlet of the coal economizer, and the adjustment coefficient increases by 0.1 for every 1% / min increase in load change rate, and decreases by 0.05 for every 0.1% / s increase in oxygen content change rate, and the final adjustment coefficient is limited to the range of 0.8-1.2. Multiplying the coordinated factor reference value by the dynamic adjustment coefficient obtains the preliminary system dynamic coordinated factor, which ensures that it can reflect the change trend of the system operation state in real time.

[0074] In step S243, the preliminary system dynamic coordinated factor is subjected to nonlinear amplitude limiting processing to ensure that its value is within the preset effective range, and the processed result is output as the final system dynamic coordinated factor.

[0075] The nonlinear amplitude limiting process uses a hyperbolic tangent function for smooth limiting, which maps the preliminary system dynamic coordination factor to the effective range of 0-1. When the input value is in the interval of 0.3-0.7, the output remains linear; when the input value is less than 0.3, the output approaches 0 but remains a minimum of 0.1; when the input value is higher than 0.7, the output approaches 1 but remains a maximum of 0.95. This processing method not only ensures the rationality of the output value, but also avoids the sudden change phenomenon at the boundary. In practical application, a change rate limit is also set to ensure that the system dynamic coordination factor does not change more than 0.2 per minute, preventing control instability caused by dramatic fluctuations in parameters. The processed system dynamic coordination factor is updated every 5 seconds and transmitted to the collaborative control system through the Modbus communication protocol as the core parameter for coordinated operation of the ammonia production and denitration systems, providing a stable and reliable decision basis for subsequent optimization control.

[0076] By establishing a stability index calculation system based on multi-parameter weighted evaluation, combining piecewise linear functions and dynamic adjustment mechanisms, and using nonlinear amplitude limiting processing, the technical solution realizes precise calculation and stable output of the system dynamic coordination factor, effectively solves the dynamic matching problem in the coordinated control of the ammonia production and denitration systems, provides a reliable decision basis for whole-process collaborative optimization, and significantly improves the adaptability and stability of the system under variable load conditions.

[0077] Further, the system dynamic coordination factor based on the load variation parameter of the boiler in step S250 cooperatively corrects the predicted value of the NO x Concentration, including: In step S251, the load variation influence degree is calculated based on the boiler load change rate, the load target tracking coefficient is calculated based on the deviation between the boiler load target value and the current load value, and the two are weighted and fused to obtain the load dynamic correction coefficient.

[0078] The calculation of the load variation influence degree is based on the absolute value of the boiler load change rate, and a piecewise function is used for processing. When the load change rate is less than 2% / min, the influence degree has a linear relationship with the change rate, and the coefficient is 0.1; when the change rate is between 2%-5% / min, the coefficient is adjusted to 0.15; when the change rate exceeds 5% / min, the coefficient is further increased to 0.2. The load target tracking coefficient is calculated by an exponential decay function, with a base value of 1.0, which decreases as the deviation between the load target value and the current load value increases, and decreases by 0.1 for every 10MW of deviation, with a minimum of 0.7. In the weighted fusion process, the weight of the load variation influence degree is 0.6, and the weight of the load target tracking coefficient is 0.4, and the two are weighted and summed to obtain the load dynamic correction coefficient. The effective range of this coefficient is controlled between 0.5-1.5, ensuring that the correction intensity can be adjusted in time when the load changes rapidly, while maintaining a moderate correction amplitude under steady-state conditions.

[0079] Step S252, the system dynamic coordination factor is used as a correction weight to weight and adjust the load dynamic correction coefficient, to obtain a coordinated correction coefficient.

[0080] When the system dynamic coordination factor is used as a correction weight to participate in calculation, it is first normalized to map it to the range of 0.8-1.2. When the coordination factor is high (greater than 0.8), it indicates that the system is running well, and at this time the adjustment of the load dynamic correction coefficient is relatively mild; when the coordination factor is low (less than 0.5), it indicates that the system has obvious mismatch, and at this time the correction effort needs to be increased. In specific implementation, the coordinated correction coefficient is obtained by the product of the load dynamic correction coefficient and the system dynamic coordination factor, and a smoothing mechanism is introduced to avoid sudden changes in the coefficient. At the same time, the coefficient change rate limit is set to ensure that the change per minute does not exceed 0.3, to maintain the stability of the control system. In special working conditions, such as rapid load change of the unit, a feedforward compensation mechanism is also introduced to pre-adjust the coordinated correction coefficient according to the load change trend, to improve the response speed of the system.

[0081] Step S253, the SCR inlet NO x concentration prediction value is multiplied by the coordinated correction coefficient to obtain the NO x concentration value after coordinated correction, and saturation limiting processing is performed to obtain the corrected SCR inlet NO x concentration prediction value.

[0082] The SCR inlet NO x concentration prediction value is multiplied by the coordinated correction coefficient, and then strict saturation limiting processing is performed. First, the upper and lower limit values based on the unit operation experience are set, the lower limit is 70% of the theoretical NO x concentration generated under the current load, and the upper limit is 130%. In the processing, a limiting strategy with a dead zone is adopted. When the corrected NO x concentration value is within the normal range (90%-110% of the theoretical value), it is directly output; when it is outside this range but within the upper and lower limits, a first-order inertial filter is used for smooth output; when it is outside the upper and lower limits, it is forcibly limited to the boundary value. At the same time, in order to ensure the reliability of the control system, an abnormal data rejection mechanism is set. When the difference between the corrected NO x concentration value and the value at the previous moment exceeds 50 mg / m³, the data validity check is started to avoid control errors caused by sensor failure or data transmission abnormalities. The final output of the corrected SCR inlet NO x concentration prediction value retains the trend characteristics of the original prediction and also incorporates real-time correction of the system operation state, providing a more accurate and reliable input for subsequent ammonia demand calculation.

[0083] Specifically, the step S300 includes the following steps. In the step S310, the boiler load change rate, the load target value and the SCR inlet NO x concentration prediction value are collected in real time to identify the working condition mode.

[0084] In the process of unit operation, the working condition mode identification system comprehensively judges the three key parameters of the boiler load change rate, the load target value and the SCR inlet NO x concentration prediction value. The load change rate reflects the dynamic adjustment demand of the unit, and when the absolute value of the change rate exceeds 0.5% / min, it is determined as a variable load working condition. The load target value is used to identify the running target state of the unit, and the load deviation can be calculated in combination with the current load value. The SCR inlet NO x concentration prediction value provides the forward-looking information of the environmental protection constraint condition. The system divides the working condition into four modes of stable state, load increase, load decrease and deep peak shaving according to the combination characteristics of the three parameters, and each mode corresponds to different control requirements. In actual application, a delay confirmation mechanism is also introduced, which requires the working condition characteristics to be maintained for more than 10 seconds before mode switching to avoid misjudgment caused by instantaneous fluctuation of parameters.

[0085] In the step S320, the control priority is dynamically determined by the multi-objective coordinator according to the identified working condition mode, and the multi-objective coordinator has an optimization strategy corresponding to each working condition mode.

[0086] The multi-objective coordinator calls the corresponding optimization strategy from the built-in expert rule base according to the identified working condition mode. In the load increase working condition, the control priority is set as the SCR denitration system priority to ensure the rapid response to the rising trend of NO x concentration. In the load decrease working condition, the control priority is adjusted to the hydrolyzer ammonia production system priority to avoid excess ammonia supply. In the stable state working condition, the balanced optimization strategy is adopted to consider the denitration efficiency and operation economy. In the deep peak shaving working condition, the main target is to improve the system adaptability. The optimization strategy is specifically realized by adjusting the control parameters, including the response speed, the control gain and the amplitude limit range. The coordinator re-evaluates the control priority every 5 seconds and fine-tunes according to the actual operation state of the unit to ensure that the control strategy is always optimally matched with the current working condition.

[0087] In the step S330, the first control signal for the SCR denitration system and the second control signal for the hydrolyzer ammonia production system are respectively generated by the cooperative control signal generator based on the dynamically determined control priority and the lead ammonia demand signal.

[0088] The cooperative control signal generator receives the control priority and the lead ammonia demand signal, and generates the control signal using a double-channel parallel processing architecture. For the SCR denitration system, the first control signal is mainly generated by a feedforward-feedback compound control algorithm. The feedforward part is based on the lead ammonia demand signal for fast response, and the feedback part is based on the NOx concentration at the SCR outlet for adjustment. The second control signal is generated by a predictive control algorithm, which takes into account the hydrolysis reaction characteristics and equipment operation constraints, and calculates the optimal control amount through material balance and energy balance. x The concentration measured value is finely adjusted. For the hydrolysis reactor ammonia production system, the second control signal uses a predictive control algorithm, which takes into account the hydrolysis reaction characteristics and equipment operation constraints, and calculates the optimal control amount through material balance and energy balance. The generation processes of the two control signals maintain strict time sequence synchronization to ensure the coordination of ammonia supply and consumption. At the same time, the control system also sets amplitude limit and change rate limit, the change rate of the first control signal is limited to ±5% / min, and the change rate of the second control signal is limited to ±3% / min, to prevent control command mutation from causing impact on the equipment.

[0089] In step S340, the ammonia injection optimization instruction is generated based on the first control signal, and the ammonia production optimization instruction is generated based on the second control signal, to realize the cooperative control of the SCR denitration system and the hydrolysis reactor ammonia production system.

[0090] The generation of the ammonia injection optimization instruction is based on the first control signal, and is realized by a cascade PID controller for precise adjustment. The main PID controller takes the NOx concentration at the SCR outlet as the controlled variable, and the auxiliary PID controller takes the ammonia injection flow as the controlled variable, to ensure the accuracy of ammonia injection through double-loop control. The generation of the ammonia production optimization instruction is more complex, which needs to decompose the second control signal into three sub-instructions: the urea solution feeding instruction is executed by a flow controller, the steam feeding instruction is adjusted by a pressure-temperature compound controller, and the product gas output instruction is managed by a pressure controller. Strict logical interlocking is maintained between the instructions, such as the urea solution feeding must be allowed to enter automatic control only after the hydrolysis reactor temperature reaches the set value. All optimization instructions are output to the field execution mechanism through 4-20mA standard signal, and instruction validity verification is set, which automatically switches to a safe control mode when detecting execution mechanism failure or feedback anomaly, to ensure system operation reliability.

[0091] By establishing an intelligent decision-making mechanism based on working condition recognition and a multi-objective coordinated optimization strategy, combined with double-channel cooperative control signal generation and fine instruction distribution, the technical scheme realizes precise cooperative control of the hydrolysis reactor ammonia production system and the SCR denitration system in all working conditions, effectively solves the response lag and coordination deficiency problems existing in traditional control methods, significantly improves the control quality and operation economy of the denitration system, and ensures the safe and stable operation of the system.

[0092] Further, in step S320, the control priority is dynamically determined by the multi-objective coordinator according to the recognized working condition mode, including: Step S321: Based on the operating condition mode recognition results, obtain the threshold values ​​of key performance indicators corresponding to the current unit operating status.

[0093] When determining control priorities, the first step is to obtain the corresponding key performance indicator (KPI) thresholds based on the identified specific operating conditions. For increased load conditions, the upper limit of ammonia slip concentration in the KPI thresholds is set at 2.7 ppm, and the SCR outlet NO... x The upper limit of concentration is set at 42 mg / Nm³, and the safety range boundary value of the hydrolyzer liquid level is set within 5% above and below the set liquid level value; for reduced load conditions, these thresholds are adjusted accordingly to an upper limit of 2.5 ppm for ammonia slip concentration and an SCR outlet NO x The concentration limit is 38 mg / Nm³, and the safe range boundary value of the hydrolyzer liquid level is set within 5% above and below the set liquid level value. These threshold settings are based on the statistical analysis results of a large amount of historical operating data, and take into account the differences in equipment operating characteristics and environmental protection requirements under different operating conditions, providing clear boundary conditions for subsequent optimization strategy selection.

[0094] Step S322: Based on the key performance indicator thresholds, select an optimization strategy that matches the current operating mode through the preset expert rule base built into the multi-objective coordinator.

[0095] Based on the acquired key performance indicator thresholds, the multi-objective coordinator selects the most suitable optimization strategy for the current operating conditions through its built-in expert rule base. The expert rule base contains dozens of practically validated control rules, such as those for situations where the unit is under increased load and the SCR outlet NO... x When the concentration approaches the upper limit, the "denitrification priority" strategy is automatically selected; when the hydrolyzer level approaches the safety boundary, the "ammonia production system protection" strategy is activated. Each rule undergoes rigorous logical verification to ensure optimal control guidance under various operating conditions. In actual operation, the rule base also has a self-learning function, which can continuously optimize rule parameters based on historical data of control effects. For example, in the application of a 1050MW unit, through three months of accumulated operating data, the rule trigger threshold under load conditions was optimized from 40mg / Nm³ to 38mg / Nm³, further improving control accuracy.

[0096] Step S323: Based on the selected optimization strategy, determine the dynamic weight parameters used for coordinated control. The dynamic weight parameters include the SCR-side control priority coefficient and the hydrolyzer-side control priority coefficient.

[0097] According to the selected optimization strategy, the dynamic weight parameters for coordinated control are determined. In the "denitration priority" strategy, the SCR side control priority coefficient is set to 0.7-0.9, and the hydrolyzer side control priority coefficient is set to 0.3-0.1 accordingly; in the "ammonia system protection" strategy, the weight distribution is reversed, the hydrolyzer side control priority coefficient is raised to 0.6-0.8, and the SCR side control priority coefficient is reduced to 0.4-0.2; in the balanced optimization strategy, the priority coefficients of both sides are kept at 0.5. These weight parameters are not fixed values, but are dynamically adjusted within a certain range according to the actual operating state, such as adding a temporary bias of 0.1-0.2 to the basic priority coefficient during rapid load change to enhance the response ability of the system. The determination of the priority coefficient also takes into account long-term factors such as equipment running time and catalyst activity, ensuring that the control strategy meets the immediate needs and also takes into account the long-term reliability of the equipment.

[0098] By establishing a key performance indicator threshold system based on operating condition characteristics, combined with an intelligent expert rule base and a dynamic priority coefficient distribution mechanism, the technical solution realizes precise and dynamic determination of control priority, ensuring that the ammonia production and denitration systems can obtain the optimal allocation of control resources under different operating conditions, effectively improving the system's ability to cope with complex conditions, while maximizing the system's operating economy and reliability while ensuring environmental protection indicators.

[0099] Further, in step S321, based on the operating condition mode recognition result, the key performance indicator threshold corresponding to the current unit operating state is obtained, including: In step S3211, according to the operating condition mode recognition result, the basic threshold of the key performance indicator is read from the pre-set operating condition-threshold mapping relationship, the basic threshold includes the basic ammonia escape concentration upper limit value, the basic SCR outlet NO x concentration upper limit value and the basic hydrolyzer liquid level safety range boundary value.

[0100] The establishment of the operating condition-threshold mapping relationship is based on the statistical analysis results of a large amount of historical operating data, and different basic threshold parameters are set for the four main operating condition modes. For the load increase condition, the basic ammonia escape concentration upper limit value is set to 2.5 ppm, the basic SCR outlet NO x concentration upper limit value is set to 45 mg / Nm³, and the basic hydrolyzer liquid level safety range boundary value is set according to the actual equipment specifications, for example, expressed in absolute liquid level value, such as 1000 mm ± 50 mm (corresponding to a fluctuation range of about ± 5%), to ensure that the liquid level is stable within the safety range; for the load reduction condition, the corresponding adjustment is ammonia escape concentration upper limit 2.0 ppm, SCR outlet NO xThe upper limit of the concentration is 40 mg / Nm3, and the safe range of the hydrolyzer liquid level is 35%-75%. In the steady state operation condition, relatively loose threshold settings are adopted, which are 3.0 ppm, 50 mg / Nm3 and 45%-85% respectively. In the deep peak regulation condition, the most stringent threshold control is adopted, which is 1.8 ppm, 35 mg / Nm3 and 50%-70% respectively. These basic threshold values are used as reference values, which will be dynamically corrected according to the specific operation state of the unit in actual application, to ensure that the environmental protection requirements are met and the equipment operation safety is considered.

[0101] In step S3212, the unit operation state parameters are monitored in real time, including the boiler load rate, the boiler load change rate, the oxygen content at the economizer outlet, and the total air volume to total coal volume ratio.

[0102] The real-time monitoring of the unit operation state parameters adopts a distributed acquisition system. The boiler load rate is obtained through the generator power signal, with a sampling frequency of 1 second. The boiler load change rate is obtained by calculating the load difference between adjacent 10 seconds, which can accurately reflect the unit operation trend. The oxygen content at the economizer outlet is measured by a high-precision zirconia oxygen analyzer, with a measurement accuracy of ±0.1%. The total air volume to total coal volume ratio is obtained by real-time calculation of the ratio of the sum of primary air volume and secondary air volume to coal supply amount, which directly reflects the combustion air distribution condition. All parameters are collected through the unit DCS system, using 4-20 mA standard signal transmission, and multiple filtering processes are set in the data acquisition process, including first-order inertia filtering and moving average filtering, to ensure the stability and reliability of the data, providing accurate input basis for subsequent threshold correction.

[0103] In step S3213, based on the real-time monitored unit operation state parameters, the corresponding threshold correction coefficient is queried through the pre-set threshold correction coefficient table, and the threshold correction coefficient table stores the correction coefficient values corresponding to different unit operation state parameter intervals.

[0104] The threshold correction coefficient table adopts a multi-dimensional query structure, and sets corresponding correction coefficients according to different operating state parameter intervals. The construction of the coefficient table is based on a large amount of historical operation data and analysis results of system dynamic characteristics, adopts a multi-dimensional query structure, and comprehensively considers the synergistic effects of multiple key operating parameters such as boiler load rate, load change rate, oxygen content at the outlet of the economizer, and air-coal ratio. The core design principle is: when the unit is running in the design optimal working condition interval, the baseline correction coefficient is assigned as 1.0, and the effectiveness of the original threshold is maintained; when the parameters deviate from the optimal interval (such as too low or too high load, rapid load change, or deviated air distribution from the optimal setting), the correction coefficient is adaptively adjusted according to the direction and degree of deviation. This adjustment follows a preset gradient rule, aiming to balance the denitration performance and system safety, for example, in the working condition that is not conducive to stable denitration or has potential risks, the system tightens the control threshold by reducing the correction coefficient, and implements a more conservative protective strategy; in the working condition that requires priority to ensure denitration efficiency, the threshold limit is relaxed by moderately increasing the correction coefficient. The adjustment of all correction coefficients is limited within a reasonable overall range, ensuring that the threshold correction behavior has both necessary flexibility and always stays within the preset safety boundary, thereby significantly improving the adaptability and robustness of the control system under different operating conditions.

[0105] In step S3214, the basic threshold value is multiplied by the corresponding threshold correction coefficient to obtain a key performance indicator threshold value corresponding to the current unit operating state.

[0106] The multiplication of the basic threshold value and the correction coefficient adopts a parameter-independent processing method. The correction of the ammonia escape concentration upper limit value mainly refers to the influence of the load rate and the load change rate, the correction of the SCR outlet NO x The correction of the ammonia escape concentration upper limit value mainly refers to the influence of the load rate and the load change rate, the correction of the SCR outlet NO

[0107] Further, the control priority based on dynamic determination in step S330 and the lead ammonia demand signal are input into a cooperative control signal generator to generate a first control signal for the SCR denitration system and a second control signal for the hydrolysis device ammonia generation system, including: In step S331, the SCR side amplitude coefficient and the hydrolysis device side amplitude coefficient are obtained based on the control priority.

[0108] The determination of the SCR-side amplitude coefficient and the hydrolyzer-side amplitude coefficient is based on the specific value of the dynamic control priority, and a linear mapping relationship is used for conversion. When the control priority is completely biased towards the SCR denitration system (the priority coefficient is 1.0), the SCR-side amplitude coefficient is set to 1.2, and the hydrolyzer-side amplitude coefficient is correspondingly set to 0.8; when the control priority is completely biased towards the hydrolyzer ammonia production system (the priority coefficient is 0), the SCR-side amplitude coefficient is adjusted to 0.8, and the hydrolyzer-side amplitude coefficient is adjusted to 1.2; in the balanced control state (the priority coefficient is 0.5), both the amplitude coefficients are kept at 1.0. In actual application, a feedforward compensation of load change trend is also introduced, and when a rapid load rise is detected, an additional temporary bias of 0.1 is added to the basic amplitude coefficient, so as to ensure that the system has sufficient adjustment capacity. The update period of the amplitude coefficient is set to 10 seconds, and a first-order inertia filter is used for processing, so as to avoid the impact of sudden changes of the coefficient on the control system.

[0109] In step S332, the SCR-side reference control signal is calculated based on the lead ammonia demand signal and the SCR-side amplitude coefficient, and the hydrolyzer-side reference control signal is calculated based on the lead ammonia demand signal and the hydrolyzer-side amplitude coefficient.

[0110] The calculation of the SCR-side reference control signal is obtained by directly multiplying the lead ammonia demand signal and the SCR-side amplitude coefficient, which fully considers the fast response characteristics of the SCR denitration system. The calculation of the hydrolyzer-side reference control signal uses a more complex processing method, in addition to multiplying the lead ammonia demand signal and the hydrolyzer-side amplitude coefficient, a compensation correction of the hydrolyzer dynamic characteristics is also introduced, including a feedforward compensation based on the hydrolyzer liquid level change rate and a feedback correction based on the hydrolyzer outlet pressure. The calculation of the two reference control signals maintains strict synchronization, ensuring that it is completed simultaneously in any calculation period. Signal validity check is set in the calculation process, when the amplitude coefficient is abnormal or the lead ammonia demand signal exceeds the reasonable range, the valid value of the last period is automatically enabled to maintain the stable operation of the control system. The calculation result of the reference control signal also undergoes preliminary amplitude limiting processing to ensure that the value is within the safe operation range of the equipment.

[0111] In step S333, the SCR-side maximum change rate limit value and the hydrolyzer-side maximum change rate limit value are obtained by querying the preset response rate configuration table according to the control priority.

[0112] The response rate configuration table sets multiple groups of rate limit parameters according to different control priorities. When the control priority is biased towards the SCR denitration system, the maximum change rate limit value on the SCR side is set to ±8% / min, and the maximum change rate limit value on the hydrolyzer side is set to ±3% / min; when the control priority is biased towards the hydrolyzer ammonia production system, the limit value on the SCR side is adjusted to ±5% / min, and the limit value on the hydrolyzer side is adjusted to ±4% / min; in the deep balancing control state, the limit values on both sides are set to ±6% / min. In addition, the rate limit value is dynamically adjusted according to the unit operation state. During the rapid load change period, the limit value on the SCR side can be temporarily relaxed to ±10% / min, and the limit value on the hydrolyzer side is tightened to ±2% / min accordingly. The query period of the configuration table is 5 seconds, and the limit value parameters are adjusted in advance through change trend prediction, ensuring that the control system not only maintains rapid response capability, but also avoids the occurrence of over-regulation phenomenon.

[0113] In step S334, the change rate limit processing of the initial control quantity on the SCR side is performed based on the maximum change rate limit value on the SCR side, and the processed result is taken as the first control signal.

[0114] The change rate limit processing of the initial control quantity on the SCR side adopts a slope limit algorithm with a dead zone. When the change rate of the control quantity is within the maximum limit value range, the original value is directly output; when the change rate exceeds the limit value, the maximum allowed change rate is output. In specific implementation, the calculation of the change rate adopts backward difference method, which compares the control quantity difference between the current period and the last period, and divides the sampling period to obtain the actual change rate. The limit algorithm also sets different limit values for acceleration and deceleration. The acceleration process limit is relatively loose (+8% / min), and the deceleration process limit is relatively strict (-6% / min), which avoids the rapid reduction of ammonia injection quantity affecting the denitration effect. The processed first control signal is subjected to smoothing filtering, a second-order Butterworth filter is used to eliminate high-frequency fluctuations, to ensure the stability of the control instruction, and finally output to the SCR ammonia injection regulating valve through a 4-20mA analog signal.

[0115] In step S335, the change rate limit processing of the initial control quantity on the hydrolyzer side is performed based on the maximum change rate limit value on the hydrolyzer side, and the processed result is taken as the second control signal.

[0116] The rate-of-change limitation processing of the initial control quantity on the hydrolyzer side takes into account the significant inertia of the ammonia production process and adopts a more stringent control strategy. The rate-of-change limitation algorithm includes dual constraints of rate limitation and acceleration limitation, limiting not only the change amplitude per minute but also the gradient of the rate of change. In the processing, the control quantity is first filtered by the setpoint, using a first-order inertial element to convert the step change into a ramp change. The filtering time constant is dynamically adjusted according to the hydrolyzer's operating status, being 30 seconds under normal conditions and extended to 60 seconds under abnormal conditions. Subsequently, rate-of-change limitation is applied, with the limit value adjusted in real time according to the hydrolyzer's liquid level and outlet pressure. When the liquid level is abnormal or the pressure fluctuates, the limit amplitude is automatically tightened. The final generated second control signal is decomposed into three sub-instructions, output to the urea solution feed valve, steam feed valve, and product gas output valve, respectively. Strict logical interlocking relationships are maintained between these sub-instructions to ensure the safe and stable operation of the hydrolyzer ammonia production system.

[0117] By establishing a dynamic adjustment mechanism for amplitude coefficients and rate limits based on control priorities, and combining it with a rate-of-change limiting algorithm for different system characteristics, this technical solution achieves precise coordinated control of the SCR denitrification system and the hydrolyzer ammonia production system. This ensures both the rapid response capability of the SCR system and the operational stability of the hydrolyzer system, effectively solving the response mismatch problem in traditional control and significantly improving the quality of the whole-process coordinated control and the reliability of system operation.

[0118] In summary, this invention constructs a NO based on an LSTM neural network. x A concentration prediction model accurately predicts the trend of nitrogen oxide concentration changes at the SCR inlet and, combined with dynamic boiler load parameters and the real-time operating status of the hydrolyzer-SCR system, calculates an advance ammonia demand signal to guide the coordinated operation of the entire process. Then, a multi-objective coordinator dynamically optimizes control priorities, and a coordinated control signal generator synchronously generates optimized instructions for ammonia injection into the SCR denitrification system and ammonia production in the hydrolyzer, ultimately achieving precise matching and advance regulation of the ammonia "production-supply-use" process. This solution effectively solves the problems of untimely ammonia injection control, supply-demand imbalance, and large operational fluctuations caused by measurement lag and system isolation in traditional control methods. It significantly improves the control quality and adaptability of the denitrification system under complex operating conditions such as deep peak shaving, achieving a comprehensive improvement in ammonia consumption reduction, catalyst life extension, and system operational safety while ensuring environmental protection indicators.

[0119] Accordingly, a second aspect of the present invention provides an electronic device, including: at least one processor; and a memory connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the above-mentioned NO-based... xPredictive hydrolyzer ammonia production and denitration whole process collaborative control method

[0120] Correspondingly, the third aspect of the embodiment of the present application provides a computer readable storage medium, which stores computer instructions, and the instructions are executed by a processor to realize the above-mentioned NO x Predictive hydrolyzer ammonia production and denitration whole process collaborative control method

[0121] The embodiment of the present application aims to protect a NO x The predictive hydrolyzer ammonia production and denitration whole process collaborative control method has the following effects: 1. By establishing an LSTM neural network prediction model based on an LSTM neural network, the response delay problem existing in traditional extraction type CEMS measurement devices is effectively overcome, the change trend of the SCR inlet NO x Concentration can be obtained in advance and accurately, which provides a key leading decision basis for subsequent ammonia demand calculation and whole process collaborative control, thereby significantly improving the timeliness and accuracy of ammonia injection control, and enabling the denitration system to still maintain stable outlet NO x Concentration under the condition of rapid load change of the unit, and avoiding the risk of environmental protection index fluctuation and exceeding due to control lag; 2. A whole process collaborative control architecture covering hydrolyzer ammonia production and SCR denitration is designed, the control priority is dynamically determined by a multi-objective coordinator, and a single leading ammonia demand signal is converted into a control instruction guiding the collaborative action of the two systems by using a collaborative control signal generator, thereby fundamentally solving the "control island" problem of independent ammonia production and denitration in traditional control, realizing accurate matching and linkage response of the ammonia "production, supply and use" links, and greatly relieving the imbalance between supply and demand and the outlet pressure fluctuation of the hydrolyzer caused by load mutation; 3. By constructing a whole process control strategy based on dynamic weight parameters and collaborative correction mechanism, adaptive collaborative optimization of the hydrolyzer ammonia production system and the SCR denitration system under different operating conditions is realized; the amplitude and response rate of the control signal can be dynamically adjusted according to real-time load changes, key performance index thresholds and system operating states, so that the SCR system with fast response and the hydrolyzer system with large inertia can be in the best cooperation state, thereby effectively reducing the ammonia consumption of the system, prolonging the service life of the catalyst, and significantly improving the operation economy and stability of the whole denitration system under the premise of ensuring the denitration efficiency.

[0122] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0123] The present application is described in reference to the flowchart and / or block diagrams of the method, apparatus (system) and computer program product according to embodiments of the application. It should be understood that each flow and / or block in the flowchart and / or block diagrams, and a combination of flows and / or blocks in the flowchart and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, a special purpose computer, an embedded processor or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagrams block or blocks. Figure 1 one or more flows and / or blocks Figure 1 an apparatus with a function specified in one or more flows and / or blocks.

[0124] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart and / or block diagrams block or blocks. Figure 1 one or more flows and / or blocks Figure 1 an apparatus with a function specified in one or more flows and / or blocks.

[0125] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart and / or block diagrams block or blocks. Figure 1 one or more flows and / or blocks Figure 1 an apparatus with a function specified in one or more flows and / or blocks.

[0126] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing and illustrating, not limiting, the technical solutions of the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A method of producing ammonia based on NO x The method for the full-process synergistic control of the predicted hydrolyzer ammonia production and denitration is characterized in that, The method comprises the following steps: Collecting real-time state parameters of the boiler operation, inputting the real-time state parameters into an LSTM neural network prediction model to obtain a predicted value of the SCR inlet NOx concentration in advance of actual measurement; Based on the SCR inlet NO x The concentration prediction value, in combination with a load variation parameter of the boiler, is used to calculate a leading ammonia demand signal, which is used to simultaneously guide the leading action of the hydrolyzer ammonia production system and the SCR denitration system. Using the advance ammonia demand signal to synchronously generate an ammonia injection optimization instruction for controlling the SCR denitration system and an ammonia production optimization instruction for controlling the hydrolyzer ammonia production system.

2. The NOx prediction-based hydrolyzer ammonia production and denitration full-process collaborative control method according to claim 1, characterized in that, The use of the advance ammonia demand signal to synchronously generate an ammonia injection optimization instruction for controlling the SCR denitration system and an ammonia production optimization instruction for controlling the hydrolyzer ammonia production system comprises: The boiler load change rate, the load target value and the SCR inlet NO x Concentration prediction value, operating mode recognition is performed on the current operating state of the unit. Dynamically determining a control priority according to the identified operating mode; Based on the dynamically determined control priority and the advance ammonia demand signal, a first control signal for the SCR denitration system and a second control signal for the hydrolyzer ammonia production system are respectively generated; Based on the first control signal, the ammonia injection optimization instruction is generated, and based on the second control signal, the ammonia production optimization instruction is generated, to realize coordinated control of the SCR denitration system and the hydrolyzer ammonia production system.

3. The NOx-based system of claim 2, wherein the NOx-based system is a NOx-based system for an internal combustion engine. x The method for the whole-process collaborative control of the predicted hydrolyzer ammonia production and denitration, characterized in that, The dynamically determining a control priority according to the identified operating mode comprises: Based on the operating mode recognition result, a key performance indicator threshold value corresponding to the current unit operating state is obtained; According to the key performance indicator threshold value, an optimization strategy matching the current operating mode is selected through a pre-set expert rule base; According to the selected optimization strategy, a dynamic weight parameter for coordinated control is determined, and the dynamic weight parameter comprises an SCR side control priority coefficient and a hydrolyzer side control priority coefficient.

4. The NOx-based system of claim 3, wherein the NOx-based system is a NOx trap system. x The method for the whole-process collaborative control of the predicted hydrolyzer ammonia production and denitration, characterized in that, The based on the operating mode recognition result, a key performance indicator threshold value corresponding to the current unit operating state is obtained, comprising: According to the working condition mode recognition result, a basic threshold of a key performance indicator is read from a preset working condition-threshold mapping relationship, the basic threshold including a basic ammonia escape concentration upper limit value, a basic SCR outlet NO x concentration upper limit value, and a basic hydrolyzer liquid level safety range boundary value; Real-time monitoring of unit operating state parameters, including boiler load rate, boiler load change rate, economizer outlet oxygen content, total air volume to total coal volume ratio; Based on the real-time monitored unit operating state parameters, a corresponding threshold correction coefficient is queried through a pre-set threshold correction coefficient table, and the threshold correction coefficient table stores correction coefficient values corresponding to different unit operating state parameter intervals; The base threshold value is multiplied by the corresponding threshold correction coefficient to obtain a key performance indicator threshold value corresponding to the current unit operating state.

5. The NOx-based system of claim 2, wherein the NOx-based system is a NOx trap system. x The method for the whole-process collaborative control of the predicted hydrolyzer ammonia production and denitration, characterized in that, The based on the dynamically determined control priority and the advance ammonia demand signal, a first control signal for the SCR denitration system and a second control signal for the hydrolyzer ammonia production system are respectively generated, comprising: Based on the control priority, an SCR side amplitude coefficient and a hydrolyzer side amplitude coefficient are obtained; Based on the advance ammonia demand signal, an SCR side reference control signal is calculated in combination with the SCR side amplitude coefficient, and a hydrolyzer side reference control signal is calculated in combination with the hydrolyzer side amplitude coefficient; According to the control priority, an SCR side maximum change rate limit value and a hydrolyzer side maximum change rate limit value are obtained by querying a pre-set response rate configuration table; The initial control amount on the SCR side is subjected to a rate limiting process according to the maximum rate limiting value on the SCR side, and the processed result is taken as the first control signal; The initial control amount on the hydrolyzer side is subjected to a rate limiting process according to the maximum rate limiting value on the hydrolyzer side, and the processed result is taken as the second control signal.

6. The NOx-based composition of any one of claims 1-5. x The method for the collaborative control of the whole process of the predicted hydrolyzer ammonia production and denitration, characterized in that, The SCR inlet NO x The concentration prediction value, in combination with a load variation parameter of the boiler, is used to calculate a lead ammonia demand signal, comprising: A first key parameter reflecting the dynamic characteristics of the hydrolyzer ammonia production system is obtained, and the first key parameter includes the hydrolyzer liquid level change rate, the real-time measured value of the hydrolyzer outlet pressure and the urea solution supply flow; A second key parameter reflecting the operating state of the SCR denitration system is obtained, and the second key parameter includes the SCR inlet flue gas temperature change rate and the real-time measured value of the pressure difference between the SCR reactor inlet and outlet; The total air volume of the boiler, the total coal volume of the boiler and the oxygen volume at the outlet of the coal economizer are obtained, and a dynamic flue gas flow parameter is calculated; The first key parameter of the hydrolyzer ammonia production system and the second key parameter of the SCR denitration system are input into a collaborative calculation model to obtain a system dynamic coordination factor; The NOx concentration prediction value is cooperatively corrected based on the load variation parameter of the boiler and the system dynamic coordination factor. x concentration prediction value is cooperatively corrected based on the load variation parameter of the boiler and the system dynamic coordination factor. NOx concentration at the inlet of the SCR based on the coordinated correction x The concentration prediction value is fused with the dynamic flue gas flow parameter to generate the lead ammonia demand signal.

7. The NOx-based system of claim 6, wherein the NOx-based system is a NOx trap system. x The method for the whole-process collaborative control of the predicted hydrolyzer ammonia production and denitration, characterized in that, The first key parameter of the hydrolyzer ammonia production system and the second key parameter of the SCR denitration system are input into a collaborative calculation model to obtain a system dynamic coordination factor, including: The hydrolyzer operating stability index is calculated based on the first key parameter, the SCR operating stability index is calculated based on the second key parameter, the ratio of the hydrolyzer operating stability index to the SCR operating stability index is calculated, and a system stability ratio is obtained; Based on the system stability ratio and the current load rate of the unit, a coordination factor reference value is calculated by a segmented linear function, the coordination factor reference value is multiplied by a dynamic adjustment coefficient to obtain a preliminary system dynamic coordination factor, wherein the dynamic adjustment coefficient is calculated according to the boiler load change rate and the oxygen volume change rate at the outlet of the coal economizer; The preliminary system dynamic coordination factor is subjected to nonlinear amplitude limiting processing to ensure that its value is within a preset effective range, and the processed result is taken as the final system dynamic coordination factor.

8. The NOx-based system of claim 6, wherein the NOx-based system is a NOx trap system. x The method for the whole-process collaborative control of the predicted hydrolyzer ammonia production and denitration, characterized in that, The NO concentration prediction value is cooperatively corrected based on the load variation parameter of the boiler and the system dynamic coordination factor. x The NO concentration prediction value is cooperatively corrected based on the load variation parameter of the boiler and the system dynamic coordination factor. A load change influence degree is calculated based on the boiler load change rate, a load target tracking coefficient is calculated based on the deviation between the boiler load target value and the current load value, and the two are weighted and fused to obtain a load dynamic correction coefficient; The system dynamic coordination factor is taken as a correction weight to weight and adjust the load dynamic correction coefficient to obtain a collaborative correction coefficient; The SCR inlet NO x concentration prediction value is multiplied by the synergistic correction coefficient to obtain a synergistically corrected NO x concentration value, and saturation limiting processing is performed to obtain the corrected SCR inlet NO x concentration prediction value.

9. The NOx-based composition of any one of claims 1-5 x The method for the collaborative control of the whole process of the predicted hydrolyzer ammonia production and denitration, characterized in that, The real-time state parameters of the collection boiler operation are collected, the real-time state parameters are input into an LSTM neural network prediction model, and the SCR inlet NO x concentration prediction value are obtained in advance of actual measurement. Real-time state parameters in the boiler operation process are collected, and the real-time state parameters include the total air volume, the total coal volume, the main steam flow, the flue gas temperature at the outlet of the furnace, the secondary air volume of each layer, the excess air volume and the oxygen concentration at the outlet of the coal economizer; The real-time state parameters are subjected to data preprocessing to form a time series data set; Based on the LSTM neural network prediction model, the NO x concentration estimate value; The estimated NOx concentration is compared with the current SCR inlet NO. x The measured concentration values ​​were weighted and fused to obtain the SCR inlet NO. x Predicted concentration values.

10. The NOx-based system of claim 9, wherein the NOx-based system is a NOx trap system. x The method for the whole-process collaborative control of the predicted hydrolyzer ammonia production and denitration, characterized in that, The LSTM neural network prediction model combines the time series data set to calculate NO x concentration estimates, comprising: The time series data set is organized into a model input sequence according to a fixed time step, and the model input sequence is input into the LSTM neural network prediction model; The model input sequence is processed by the encoder of the LSTM neural network prediction model step by step, and each LSTM unit of the encoder calculates the hidden state at the current time according to the input at the current time and the hidden state at the previous time. inputting the hidden state of the last time step of the encoder as a context vector into a decoder of the LSTM neural network prediction model, calculating the NO x concentration estimate.

Images