Method and apparatus for predicting inlet NOX concentration of denitration system of thermal power plant

Abstract

Disclosed in the present application is a method for predicting the inlet NOx concentration of a denitration system of a thermal power plant in consideration of measurement delay. The method includes: determining influencing factors of the inlet NOx concentration of a denitration reactor, and collecting the inlet NOx concentration of the denitration reactor and data of the influencing factors of the inlet NOx concentration; determining a measurement delay time of the inlet NOx concentration, and correcting the inlet NOx concentration; calculating a correlation coefficient between the inlet NOx concentration and each of the influencing factors by using a Fourier transform correlation coefficient method, and determining an output variable and an input feature parameter variable that are used for prediction; on the basis of the determined output variable and input feature parameter variable, training a prediction model by using a machine learning algorithm based on physical constraints; and collecting data of the influencing factors of the inlet NOx concentration at a current moment, and inputting same into the prediction model to obtain an inlet NOx concentration value at the current moment. Further disclosed is an apparatus for predicting the inlet NOx concentration of a denitration system of a thermal power plant in consideration of measurement delay.

Description

Methods and devices for predicting NOx concentration at the inlet of denitrification systems in thermal power plants

[0001] This application claims priority to Chinese Patent Application No. 2025100165801, filed on January 6, 2025, entitled “Method and Apparatus for Predicting NOx Concentration at the Inlet of a Denitrification System in a Thermal Power Plant”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This document relates to the field of flue gas denitrification technology in coal-fired power plants, and in particular to a method and device for predicting NOx concentration at the inlet of a denitrification system in a thermal power plant that takes into account measurement delay. Background Technology

[0003] With the promulgation of a series of national policies and regulations, air pollutant emissions from coal-fired power plants have been subject to strict supervision. The combustion of coal generates a large amount of air pollutants, causing serious environmental pollution problems. NOx is one of the main pollutants emitted by coal-fired power plants. Once released into the atmosphere, it undergoes a series of physicochemical reactions, generating various harmful substances that cause significant damage to the environment and human health. Accurate measurement of inlet NOx concentration is crucial for effectively controlling outlet NOx emissions, and accurate prediction of real-time inlet NOx values ​​is key to improving the control effect of denitrification systems.

[0004] Currently, both domestically and internationally, NOx components in flue gas are mainly measured in real time using Continuous Emission Monitoring Systems (CEMS). However, this measurement method has drawbacks such as long data analysis time and significant lag in measurement feedback, leading to inaccurate inlet NOx measurement results. To address this, many scholars have conducted research on predicting NOx concentrations at the denitrification inlet, often employing mechanistic models (i.e., constructing physical models for calculation) or data-driven methods. However, these methods have certain problems. For example, mechanistic models are complex and lengthy to build, and parameter definition is difficult. Data-driven models are entirely based on data and are completely limited by the collected operational data. On the one hand, when the data contains noise, the model's generalization ability is poor; on the other hand, the data is based on CEMS measurements, which have a certain delay, leading to time-series mismatches in the selected variable data at the same time, inevitably resulting in inaccurate inlet NOx concentration predictions based on delayed data.

[0005] Therefore, in the process of predicting NOx concentration at the inlet, it is necessary to consider both the mechanistic model and the data, while also removing delays in the data. Thus, considering all the above factors, predicting NOx concentration at the denitrification inlet is an urgent problem to be solved in the denitrification control process of thermal power plants. Summary of the Invention

[0006] The purpose of this invention is to provide a method and apparatus for predicting NOx concentration at the inlet of a thermal power plant denitrification system, taking into account measurement delay, in order to solve the above-mentioned problems in the prior art.

[0007] This invention provides a method for predicting NOx concentration at the inlet of a denitrification system in a thermal power plant, taking into account measurement delay, comprising:

[0008] Step 101: Determine the influencing factors of NOx concentration at the inlet of the denitrification reactor, and collect data on the inlet NOx concentration and the influencing factors of the inlet NOx concentration of the denitrification reactor;

[0009] Step 102: Determine the measurement delay time of the inlet NOx concentration, and correct the inlet NOx concentration according to the measurement delay time to obtain the inlet NOx concentration without delay;

[0010] Step 103: Calculate the correlation coefficient between the inlet NOx concentration and each influencing factor using the Fourier transform correlation coefficient method. Obtain the delay time between each influencing factor and the inlet NOx concentration based on the correlation coefficient. Correct each influencing factor based on the delay time to obtain inlet NOx concentration influencing factor data without delay. Determine the output variable and input feature parameter variable used for prediction.

[0011] Step 104: Based on the no-delay inlet NOx concentration and the no-delay inlet NOx concentration influencing factor data, and according to the determined output variable and input feature parameter variable, the prediction model is trained using a physical constraint-based machine learning algorithm.

[0012] Step 105: Collect data on factors affecting the inlet NOx concentration at the current moment, correct the data based on the calculated delay time, and input the data into the prediction model to obtain the inlet NOx concentration value at the current moment.

[0013] This invention provides a device for predicting NOx concentration at the inlet of a denitrification system in a thermal power plant, taking into account measurement delay, comprising:

[0014] The data acquisition module is used to determine the influencing factors of the NOx concentration at the inlet of the denitrification reactor, and to collect data on the NOx concentration at the inlet of the denitrification reactor and the influencing factors of the NOx concentration at the inlet.

[0015] A calibration module is used to determine the measurement delay time of the inlet NOx concentration and to correct the inlet NOx concentration according to the measurement delay time to obtain an inlet NOx concentration without delay.

[0016] The calculation module is used to calculate the correlation coefficient between the inlet NOx concentration and each influencing factor using the Fourier transform correlation coefficient method, obtain the delay time between each influencing factor and the inlet NOx concentration based on the correlation coefficient, correct each influencing factor based on the delay time, obtain inlet NOx concentration influencing factor data without delay, and determine the output variable and input feature parameter variable used for prediction.

[0017] The prediction model module is used to train the prediction model using a physical constraint-based machine learning algorithm based on the no-delay inlet NOx concentration and the no-delay inlet NOx concentration influencing factors data, according to the determined output variables and input feature parameter variables; it collects the inlet NOx concentration influencing factors data at the current moment, corrects it according to the calculated delay time, and inputs it into the prediction model to obtain the inlet NOx concentration value at the current moment.

[0018] This invention also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the above-described method for predicting the NOx concentration at the inlet of a thermal power plant denitrification system, taking into account measurement delay.

[0019] This invention also provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, implements the steps of the above-described method for predicting the NOx concentration at the inlet of a thermal power plant denitrification system, taking into account measurement delay.

[0020] The embodiments of this invention consider both the mechanistic model and the operational data of factors affecting inlet NOx concentration during the prediction process. It also analyzes the delay time of inlet NOx concentration measurement itself and the delay time between inlet NOx concentration and influencing factors. The established model improves the accuracy of inlet NOx modeling and has guiding significance for reducing pollutant emissions and costs of coal-fired power units. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 is a flowchart of a method for predicting NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay, according to an embodiment of the present invention.

[0023] Figure 2 is a schematic diagram of the process of constructing a machine learning algorithm based on physical constraints according to an embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of the inlet NOx concentration prediction process under all operating conditions according to an embodiment of the present invention.

[0025] Figure 4 is a schematic diagram of a NOx concentration prediction device at the inlet of a thermal power plant denitrification system considering measurement delay according to an embodiment of the present invention.

[0026] Figure 5 is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0028] Method Implementation Examples

[0029] According to an embodiment of the present invention, a method for predicting the NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay is provided. Figure 1 is a flowchart of the method for predicting the NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay according to an embodiment of the present invention. As shown in Figure 1, the method for predicting the NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay according to an embodiment of the present invention specifically includes:

[0030] Step S101: Determine the influencing factors of NOx concentration at the inlet of the denitrification reactor, and collect data on the inlet NOx concentration and the influencing factors of the inlet NOx concentration; specifically including:

[0031] Based on the NOx formation mechanism of flue gas in thermal power plants, the factors affecting the inlet NOx concentration include: unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal pulverizing volume.

[0032] At predetermined sampling intervals, data on the inlet NOx concentration and influencing factors of the denitrification reactor over a predetermined period of time are collected from the DCS or SIS system of the thermal power plant.

[0033] Step S102 involves determining the measurement delay time of the inlet NOx concentration and correcting the inlet NOx concentration based on the measurement delay time to obtain the inlet NOx concentration without delay; specifically including:

[0034] The measurement delay time t1 is determined by CEMS measurement based on the inlet NOx concentration. This includes the gas sampling system delay time t_sampling, the transmission pipeline delay time t_transmission, and the analyzer response time t_response. The gas sampling system delay time t_sampling is caused by the delays of the sampling pipeline, sampling pump, and filter, and is determined by the ratio of the sampling pipeline length L1 to the sampling pump flow rate V1, i.e., t_sampling = L1 / V1. The transmission pipeline delay time t_transmission is the delay in the transmission of the measurement signal from the sampling point to the CEMS instrument, and is determined by the signal transmission distance L2 and the transmission rate V2, i.e., t_transmission = L2 / V2. The analyzer response time t_response is the time required for the CEMS analyzer to respond to changes in the input signal.

[0035] The measurement delay time t1 is calculated according to Formula 1. The inlet NOx concentration data collected at the current time t2 is corrected, that is, the inlet NOx concentration value at time t2 is corrected to the inlet NOx concentration value at time (t2-t1): t1=t_sampling+t_transmission+t_response Formula 1.

[0036] Step S103: The correlation coefficient between the inlet NOx concentration and each influencing factor is calculated using the Fourier transform correlation coefficient method. Based on the correlation coefficient, the delay time between each influencing factor and the inlet NOx concentration is obtained. The influencing factors are then corrected based on the delay time to obtain inlet NOx concentration influencing factor data without delay. The output variables and input feature parameter variables used for prediction are then determined. Specifically, this includes:

[0037] Let the inlet NOx concentration factor be X(t), and the influencing factors of the inlet NOx concentration be Y(t) = {y1(t), y2(t), y3(t), y4(t), y5(t), y6(t)}, where y1, y2, y3, y4, y5, y6 are the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and pulverizing coal volume, respectively; t is the index number, t = 1, 2, ..., N, and N is the number of inlet NOx concentration samples collected. Then, the Fourier cross-correlation coefficient is determined according to Formula 2: C XY (t)=F^{-1}{X(f)*(Y*(f))} Formula 2;

[0038] Among them, C XY(t) is the Fourier cross-correlation coefficient of X(t) and Y(t); F^{-1} is the inverse Fourier transform; X(f) and Y(f) are the Fourier transforms of X(t) and Y(t); Y*(f) is the complex conjugate of Y(t); * is the conjugate of a complex number;

[0039] Calculate the Fourier correlation coefficients of X(t) and Y(t) from time 0 to t3, and calculate the times t4, t5, t6, t7, t8, and t9 corresponding to the maximum values ​​of the Fourier correlation coefficients of y1, y2, y3, y4, y5, and y6. These times are the delay time between the inlet NOx concentration and its influencing factors.

[0040] Based on the calculated delay times t4, t5, t6, t7, t8, and t9, the data on factors affecting the inlet NOx concentration collected at the current time T are corrected. That is, the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at time T are corrected to the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at times (T-t4), (T-t5), (T-t6), (T-t7), (T-t8), and (T-t9), respectively.

[0041] The output variable used for prediction is determined to be the corrected inlet NOx concentration value, and the input variables are the corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume.

[0042] Step S104: Based on the no-delay inlet NOx concentration and the no-delay inlet NOx concentration influencing factor data, and according to the determined output variable and input feature parameter variable, a physical constraint-based machine learning algorithm is used to train the prediction model; specifically including:

[0043] The prediction model was constructed using the Zeldovich mechanism model based on factors such as temperature, oxygen concentration, nitrogen concentration, pressure, reaction time, and activation energy.

[0044] A backpropagation (BP) neural network model is used as the machine learning algorithm. This BP neural network model consists of three layers: an input layer, a hidden layer, and an output layer. The input layer contains six input variables: corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume. The output layer contains the corrected inlet NOx concentration value. The parameters of the BP neural network model are the weights and thresholds between the input and hidden layers, and between the hidden and output layers. The correction of these weights and thresholds is achieved through a global error function, defined by formulas 3-5.

[0045] E(θ)=Data(θ)+λPhysics(θ) Formula 3;

[0046] Where E(θ) is the global error function, Data(θ) is the BP neural network model error function, Physics(θ) is the physical model error function, and λ is the weight coefficient of the physical constraint term, which is obtained empirically or through a genetic optimization algorithm; y i This represents the actual NOx concentration at the inlet. The inlet NOx concentration is the predicted value by the BP neural network model; The inlet NOx concentration is a physical model prediction.

[0047] By minimizing E(θ), the optimal weights and thresholds between the input layer and the hidden layer, and between the hidden layer and the output layer, are obtained, thereby determining the prediction model with the optimal physical constraints.

[0048] Step S105: Collect data on factors influencing the inlet NOx concentration at the current moment, correct them according to the calculated delay times, and input them into the prediction model to obtain the inlet NOx concentration value at the current moment. Collect data on factors influencing the inlet NOx concentration at the current moment, correct them according to the calculated delay times t4, t5, t6, t7, t8, and t9, and input them into the prediction model with optimal physical constraints to obtain the inlet NOx concentration value at the current moment.

[0049] According to specific embodiments provided by the present invention, the technical effects of the embodiments of the present invention are as follows:

[0050] (1) This invention provides a method for calculating the measurement delay of NOx concentration at the inlet, and from the perspective of measurement principle, it provides a method for eliminating the measurement delay of NOx concentration, and provides a more accurate real-time measurement value of NOx concentration at the inlet, which is better able to match the real-time NOx concentration status at the inlet.

[0051] (2) This invention also provides a method for calculating the delay between inlet NOx concentration and its influencing factors based on the Fourier correlation coefficient method. By transforming time-domain data to frequency-domain data, the data characteristics at different frequencies are obtained, further highlighting the data characteristics. This technique eliminates the delay between inlet NOx concentration and its influencing factors, making the established model more consistent and accurate.

[0052] (3) This invention provides a machine learning algorithm based on physical constraints, which comprehensively considers both the physical model and the machine learning algorithm model. It takes into account both the running data and the generation mechanism, avoiding the machine learning model from ignoring physical laws due to overfitting the training data, thereby improving the generalization ability and accuracy of the model, especially in the case of insufficient data or high noise. The established model has higher accuracy and the prediction results are more consistent with reality.

[0053] The technical solutions of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] As shown in Figure 1, the specific processing includes the following:

[0055] Step S101: Based on the NOx formation mechanism of flue gas from thermal power plants, determine the influencing factors of NOx concentration at the inlet of the denitrification reactor, and collect data on NOx concentration at the inlet of the denitrification reactor and the influencing factors of NOx concentration at the inlet.

[0056] Step S201: Based on the measurement principle of NOx concentration at the inlet of the denitrification reactor, determine the measurement delay time of the inlet NOx concentration, and correct the inlet NOx concentration accordingly to obtain inlet NOx concentration data without delay.

[0057] Step S301: The correlation coefficient between the inlet NOx concentration and each influencing factor is analyzed using the Fourier transform correlation coefficient method. Based on the calculated correlation coefficient, the delay time between each influencing factor and the inlet NOx concentration is obtained. The influencing factors are then corrected to obtain the inlet NOx concentration influencing factors without delay. In this way, the output variables and input characteristic parameter variables used for prediction are determined.

[0058] Step S401, the prediction model adopts a machine learning algorithm based on physical constraints;

[0059] Step S501: Collect the factors affecting the inlet NOx concentration at the current moment, correct them according to the calculated delay time, and input them into the machine learning algorithm model based on physical constraints to obtain the inlet NOx concentration value at the current moment.

[0060] Specifically, taking a 600MW subcritical pressure boiler unit in China as an example, based on the NOx formation mechanism analysis of flue gas in thermal power plants, the main NOx influencing factors include unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume.

[0061] Specifically, the NOx concentration and its influencing factors data collected at the inlet are mainly obtained from the DCS or SIS systems of thermal power plants, based on the past year's operating data, with a sampling interval of 1 minute, resulting in approximately 525,600 sample data points.

[0062] Specifically, according to the CEMS measurement equipment used in this 600MW unit, the measurement delay time t1 is approximately 62.3s, and the inlet NOx concentration is corrected based on the t1 value.

[0063] Specifically, the delay time calculation of the factors affecting the inlet NOx concentration uses the correlation coefficient method of Fourier transform to obtain the correlation coefficients between the inlet NOx concentration and the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding rate within 0-20s (t3 is taken as 20s). Through calculation, the moments with the largest correlation coefficients for the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding rate within 0-20s are 15s, 8s, 12s, 5s, 9s, and 18s, respectively, and the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding rate are corrected based on these moments.

[0064] Referring to Figure 2, specifically, the physical constraint-based machine learning algorithm mainly adds physical model constraints to the loss function of the machine learning algorithm.

[0065] The physical model constraints are mainly constructed using the Zeldovich mechanism model, which is based on factors such as temperature, oxygen concentration, nitrogen concentration, pressure, reaction time, and activation energy.

[0066] The machine learning algorithm primarily employs a backpropagation (BP) neural network. The BP neural network model consists of three layers: an input layer, a hidden layer, and an output layer. The input layer contains six input variables: corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume. The output layer contains the corrected inlet NOx concentration value. The parameters of the BP neural network are the weights and thresholds between the input and hidden layers, and between the hidden and output layers. The correction of these weights and thresholds is mainly achieved through a global error function, defined as: E(θ)=Data(θ)+λPhysics(θ) (Equation 3)

[0067] In the formula, E(θ) is the global error function, Data(θ) is the BP neural network model error function, Physics(θ) is the physical model error function, and λ is the weight coefficient of the physical constraint term, which can be obtained by empirical method or by genetic optimization algorithm.

[0068] In the formula, y i This represents the actual NOx concentration at the inlet. The inlet NOx concentration is the predicted value by the BP neural network model.

[0069] In the formula, y i This represents the actual NOx concentration at the inlet. The inlet NOx concentration is a physical model prediction.

[0070] By minimizing E(θ), the optimal weights and thresholds between the input layer and the hidden layer, and between the hidden layer and the output layer, are obtained, thereby determining the BP neural network model with optimal physical constraints.

[0071] Specifically, the factors affecting the inlet NOx concentration at the current moment are collected, and corrections are made based on the calculated delay times t4, t5, t6, t7, t8, and t9. The results are then input into the BP neural network model with optimal physical constraints to obtain the inlet NOx concentration value at the current moment.

[0072] Referring to Figure 3, this invention can also construct a prediction model for NOx concentration at the denitrification inlet under all operating conditions. Considering that the actual unit operation includes load increase, load decrease, and stable load conditions, it is necessary to comprehensively consider all three conditions in order to construct a prediction model under all operating conditions. Specifically, in the data collection, data under the three operating conditions of load increase, load decrease, and stable load are collected respectively. Then, BP neural network models under physical constraints are established for each of the three operating conditions. The establishment process is the same as in Example 1. Example 2 provides multiple operating condition models.

[0073] When predicting the inlet NOx concentration at the current moment, it is necessary to collect the unit load data 5 minutes before and after that time period, determine whether the unit is currently in a load increase, load decrease, or stable load phase, and input the data into the corresponding model to obtain the predicted value of the inlet NOx concentration at the current moment.

[0074] This invention provides a method for predicting NOx concentration at the inlet of a thermal power plant denitrification system, taking into account measurement delay. The method accurately calculates the delay time of the auxiliary variables used in inlet NOx modeling relative to the inlet NOx, solving the problem of time-series mismatch between the selected variable data of inlet NOx and its influencing factors at the same time, eliminating the influence of pure delay, and improving the accuracy of inlet NOx modeling. This has guiding significance for reducing pollutant emissions and costs in coal-fired power units.

[0075] Device Example 1

[0076] According to an embodiment of the present invention, a device for predicting the NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay is provided. Figure 4 is a schematic diagram of the device for predicting the NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay according to an embodiment of the present invention. As shown in Figure 4, the device for predicting the NOx concentration at the inlet of a thermal power plant denitrification system considering measurement delay according to an embodiment of the present invention specifically includes:

[0077] The data acquisition module 40 is used to determine the influencing factors of the NOx concentration at the inlet of the denitrification reactor, and to collect data on the inlet NOx concentration and the influencing factors of the inlet NOx concentration; specifically, it is used for:

[0078] Based on the NOx formation mechanism of flue gas in thermal power plants, the factors affecting the inlet NOx concentration include: unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal pulverizing volume.

[0079] At predetermined sampling intervals, data on the inlet NOx concentration and influencing factors of the denitrification reactor over a predetermined period of time are collected from the DCS or SIS system of the thermal power plant.

[0080] The calibration module 42 is used to determine the measurement delay time of the inlet NOx concentration and correct the inlet NOx concentration according to the measurement delay time to obtain the inlet NOx concentration without delay; specifically, it is used for:

[0081] The measurement delay time t1 is determined by CEMS measurement based on the inlet NOx concentration. This includes the gas sampling system delay time t_sampling, the transmission pipeline delay time t_transmission, and the analyzer response time t_response. The gas sampling system delay time t_sampling is caused by the delays of the sampling pipeline, sampling pump, and filter, and is determined by the ratio of the sampling pipeline length L1 to the sampling pump flow rate V1, i.e., t_sampling = L1 / V1. The transmission pipeline delay time t_transmission is the delay in the transmission of the measurement signal from the sampling point to the CEMS instrument, and is determined by the signal transmission distance L2 and the transmission rate V2, i.e., t_transmission = L2 / V2. The analyzer response time t_response is the time required for the CEMS analyzer to respond to changes in the input signal.

[0082] The measurement delay time t1 is calculated according to Formula 1. The inlet NOx concentration data collected at the current time t2 is corrected, that is, the inlet NOx concentration value at time t2 is corrected to the inlet NOx concentration value at time (t2-t1): t1=t_sampling+t_transmission+t_response Formula 1;

[0083] Calculation module 44 is used to calculate the correlation coefficient between the inlet NOx concentration and each influencing factor using the Fourier transform correlation coefficient method, obtain the delay time between each influencing factor and the inlet NOx concentration based on the correlation coefficient, correct each influencing factor based on the delay time, obtain inlet NOx concentration influencing factor data without delay, and determine the output variable and input feature parameter variable used for prediction; specifically used for:

[0084] Let the inlet NOx concentration factor be X(t), and the influencing factors of the inlet NOx concentration be Y(t) = {y1(t), y2(t), y3(t), y4(t), y5(t), y6(t)}, where y1, y2, y3, y4, y5, y6 are the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and pulverizing coal volume, respectively; t is the index number, t = 1, 2, ..., N, and N is the number of inlet NOx concentration samples collected. Then, the Fourier cross-correlation coefficient is determined according to Formula 2: C XY (t)=F^{-1}{X(f)*(Y*(f))} Formula 2;

[0085] Among them, C XY (t) is the Fourier cross-correlation coefficient of X(t) and Y(t); F^{-1} is the inverse Fourier transform; X(f) and Y(f) are the Fourier transforms of X(t) and Y(t); Y*(f) is the complex conjugate of Y(t); * is the conjugate of a complex number;

[0086] Calculate the Fourier correlation coefficients of X(t) and Y(t) from time 0 to t3, and calculate the times t4, t5, t6, t7, t8, and t9 corresponding to the maximum values ​​of the Fourier correlation coefficients of y1, y2, y3, y4, y5, and y6. These times are the delay time between the inlet NOx concentration and its influencing factors.

[0087] Based on the calculated delay times t4, t5, t6, t7, t8, and t9, the data on factors affecting the inlet NOx concentration collected at the current time T are corrected. That is, the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at time T are corrected to the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at times (T-t4), (T-t5), (T-t6), (T-t7), (T-t8), and (T-t9), respectively.

[0088] The output variable used for prediction is determined to be the corrected inlet NOx concentration value, and the input variables are the corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume.

[0089] The prediction model module 46 is used to train the prediction model using a physical constraint-based machine learning algorithm, based on the no-delay inlet NOx concentration and the no-delay inlet NOx concentration influencing factor data, according to the determined output variable and the input feature parameter variable; it also collects the inlet NOx concentration influencing factor data at the current moment, corrects it according to the calculated delay time, and inputs it into the prediction model to obtain the inlet NOx concentration value at the current moment. Specifically, it is used for:

[0090] The prediction model was constructed using the Zeldovich mechanism model based on factors such as temperature, oxygen concentration, nitrogen concentration, pressure, reaction time, and activation energy.

[0091] A BP neural network model is used as the machine learning algorithm. The BP neural network model consists of three layers: an input layer, a hidden layer, and an output layer. The input layer contains six input variables: the corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume. The output layer contains the corrected inlet NOx concentration value. The parameters of the BP neural network model are the weights and thresholds between the input layer and the hidden layer, and between the hidden layer and the output layer. The correction of the above weights and thresholds is performed through a global error function, which is defined by formulas 3-5: E(θ)=Data(θ)+λPhysics(θ) Formula 3;

[0092] Where E(θ) is the global error function, Data(θ) is the BP neural network model error function, Physics(θ) is the physical model error function, and λ is the weight coefficient of the physical constraint term, which is obtained empirically or through a genetic optimization algorithm; y i This represents the actual NOx concentration at the inlet. The inlet NOx concentration is the predicted value by the BP neural network model; The inlet NOx concentration is a physical model prediction.

[0093] By minimizing E(θ), the optimal weights and thresholds between the input layer and the hidden layer, and the optimal weights and thresholds between the hidden layer and the output layer are obtained, thereby determining the prediction model with the optimal physical constraints.

[0094] Data on factors influencing the inlet NOx concentration at the current moment are collected, and corrections are made based on the calculated delay times t4, t5, t6, t7, t8, and t9. The data are then input into the prediction model with optimal physical constraints to obtain the inlet NOx concentration value at the current moment.

[0095] The embodiments of the present invention are device embodiments corresponding to the above method embodiments. The specific operation of each module can be understood with reference to the description of the method embodiments, and will not be repeated here.

[0096] Device Example 2

[0097] An embodiment of the present invention provides an electronic device, as shown in FIG5, including: a memory 50, a processor 52, and a computer program stored in the memory 50 and executable on the processor 52. When the computer program is executed by the processor 52, it implements the steps described in the method embodiment.

[0098] Device Example 3

[0099] This invention provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor 52, performs the steps described in the method embodiment.

[0100] The computer-readable storage media described in this embodiment include, but are not limited to, ROM, RAM, disk, or optical disk.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting NOx concentration at the inlet of a denitrification system in a thermal power plant, considering measurement delay, characterized in that, include: Step 101: Determine the influencing factors of NOx concentration at the inlet of the denitrification reactor, and collect data on the inlet NOx concentration and the influencing factors of the inlet NOx concentration of the denitrification reactor; Step 102: Determine the measurement delay time of the inlet NOx concentration, and correct the inlet NOx concentration according to the measurement delay time to obtain the inlet NOx concentration without delay; Step 103: Calculate the correlation coefficient between the inlet NOx concentration and each influencing factor using the Fourier transform correlation coefficient method. Obtain the delay time between each influencing factor and the inlet NOx concentration based on the correlation coefficient. Correct each influencing factor based on the delay time to obtain inlet NOx concentration influencing factor data without delay. Determine the output variable and input feature parameter variable used for prediction. Step 104: Based on the no-delay inlet NOx concentration and the no-delay inlet NOx concentration influencing factor data, and according to the determined output variable and input feature parameter variable, the prediction model is trained using a physical constraint-based machine learning algorithm. Step 105: Collect data on factors affecting the inlet NOx concentration at the current moment, correct the data based on the calculated delay time, and input the data into the prediction model to obtain the inlet NOx concentration value at the current moment.

2. The method according to claim 1, characterized in that, The factors influencing the NOx concentration at the inlet of the denitrification reactor were determined, and data on the inlet NOx concentration and its influencing factors were collected. Specifically, this included: Based on the NOx formation mechanism of flue gas in thermal power plants, the factors affecting the inlet NOx concentration include: unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal pulverizing volume. At predetermined sampling intervals, data on the inlet NOx concentration and influencing factors of the denitrification reactor over a predetermined period of time are collected from the DCS or SIS system of the thermal power plant.

3. The method according to claim 1, characterized in that, Determining the measurement delay time of the inlet NOx concentration and correcting the inlet NOx concentration based on the measurement delay time to obtain a delay-free inlet NOx concentration specifically includes: The measurement delay time t1 is determined by CEMS measurement based on the inlet NOx concentration. This includes the gas sampling system delay time t_sampling, the transmission pipeline delay time t_transmission, and the analyzer response time t_response. The gas sampling system delay time t_sampling is caused by the delays of the sampling pipeline, sampling pump, and filter, and is determined by the ratio of the sampling pipeline length L1 to the sampling pump flow rate V1, i.e., t_sampling = L1 / V1. The transmission pipeline delay time t_transmission is the delay in the transmission of the measurement signal from the sampling point to the CEMS instrument, and is determined by the signal transmission distance L2 and the transmission rate V2, i.e., t_transmission = L2 / V2. The analyzer response time t_response is the time required for the CEMS analyzer to respond to changes in the input signal. The measurement delay time t1 is calculated according to Formula 1. The inlet NOx concentration data collected at the current time t2 is then corrected, that is, the inlet NOx concentration value at time t2 is corrected to the inlet NOx concentration value at time (t2-t1): t1=t_sampling+t_transmission+t_response Formula 1.

4. The method according to claim 1, characterized in that, Step 103 specifically includes: Let the inlet NOx concentration factor be X(t), and the influencing factors of the inlet NOx concentration be Y(t) = {y1(t), y2(t), y3(t), y4(t), y5(t), y6(t)}, where y1, y2, y3, y4, y5, y6 are the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and pulverizing coal volume, respectively; t is the index number, t = 1, 2, ..., N, and N is the number of inlet NOx concentration samples collected. Then, the Fourier cross-correlation coefficient is determined according to Formula 2: C XY (t)=F^{-1}{X(f)*(Y*(f))} Formula 2; Among them, C XY (t) is the Fourier cross-correlation coefficient of X(t) and Y(t); F^{-1} is the inverse Fourier transform; X(f) and Y(f) are the Fourier transforms of X(t) and Y(t); Y*(f) is the complex conjugate of Y(t); * is the conjugate of a complex number; Calculate the Fourier correlation coefficients of X(t) and Y(t) from time 0 to t3, and calculate the times t4, t5, t6, t7, t8, and t9 corresponding to the maximum values ​​of the Fourier correlation coefficients of y1, y2, y3, y4, y5, and y6. These times are the delay time between the inlet NOx concentration and its influencing factors. Based on the calculated delay times t4, t5, t6, t7, t8, and t9, the data on factors affecting the inlet NOx concentration collected at the current time T are corrected. That is, the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at time T are corrected to the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at times (T-t4), (T-t5), (T-t6), (T-t7), (T-t8), and (T-t9), respectively. The output variable used for prediction is determined to be the corrected inlet NOx concentration value, and the input variables are the corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume.

5. The method according to claim 1, characterized in that, Step 104 specifically includes: The prediction model was constructed using the Zeldovich mechanism model based on factors such as temperature, oxygen concentration, nitrogen concentration, pressure, reaction time, and activation energy. A backpropagation (BP) neural network model is used as the machine learning algorithm. This BP neural network model consists of three layers: an input layer, a hidden layer, and an output layer. The input layer contains six input variables: corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume. The output layer contains the corrected inlet NOx concentration value. The parameters of the BP neural network model are the weights and thresholds between the input and hidden layers, and between the hidden and output layers. The correction of these weights and thresholds is achieved through a global error function, defined by formulas 3-5. E(θ) = Data(θ) + λPhysics(θ) Equation 3; Where E(θ) is the global error function, Data(θ) is the BP neural network model error function, Physics(θ) is the physical model error function, and λ is the weight coefficient of the physical constraint term, which is obtained empirically or through a genetic optimization algorithm; y i This represents the actual NOx concentration at the inlet. The inlet NOx concentration is the predicted value by the BP neural network model; The inlet NOx concentration is a physical model prediction. By minimizing E(θ), the optimal weights and thresholds between the input layer and the hidden layer, and between the hidden layer and the output layer, are obtained, thereby determining the prediction model with the optimal physical constraints.

6. The method according to claim 5, characterized in that, Step 105 specifically includes: Data on factors influencing the inlet NOx concentration at the current moment are collected, and corrections are made based on the calculated delay times t4, t5, t6, t7, t8, and t9. The data are then input into the prediction model with optimal physical constraints to obtain the inlet NOx concentration value at the current moment.

7. A device for predicting NOx concentration at the inlet of a denitrification system in a thermal power plant, taking into account measurement delay, characterized in that, include: The data acquisition module is used to determine the influencing factors of the NOx concentration at the inlet of the denitrification reactor, and to collect data on the NOx concentration at the inlet of the denitrification reactor and the influencing factors of the NOx concentration at the inlet. A calibration module is used to determine the measurement delay time of the inlet NOx concentration and to correct the inlet NOx concentration according to the measurement delay time to obtain an inlet NOx concentration without delay. The calculation module is used to calculate the correlation coefficient between the inlet NOx concentration and each influencing factor using the Fourier transform correlation coefficient method, obtain the delay time between each influencing factor and the inlet NOx concentration based on the correlation coefficient, correct each influencing factor based on the delay time, obtain inlet NOx concentration influencing factor data without delay, and determine the output variable and input feature parameter variable used for prediction. The prediction model module is used to train the prediction model using a physical constraint-based machine learning algorithm based on the no-delay inlet NOx concentration and the no-delay inlet NOx concentration influencing factors data, according to the determined output variables and input feature parameter variables; it collects the inlet NOx concentration influencing factors data at the current moment, corrects it according to the calculated delay time, and inputs it into the prediction model to obtain the inlet NOx concentration value at the current moment.

8. The apparatus according to claim 7, characterized in that, The acquisition module is specifically used for: Based on the NOx formation mechanism of flue gas in thermal power plants, the factors affecting the inlet NOx concentration include: unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal pulverizing volume. Data on inlet NOx concentration and influencing factors of inlet NOx concentration of the denitrification reactor were collected from the DCS or SIS system of the thermal power plant at predetermined sampling intervals over a predetermined period of time. The correction module is specifically used for: The measurement delay time t1 is determined by CEMS measurement based on the inlet NOx concentration. This includes the gas sampling system delay time t_sampling, the transmission pipeline delay time t_transmission, and the analyzer response time t_response. The gas sampling system delay time t_sampling is caused by the delays of the sampling pipeline, sampling pump, and filter, and is determined by the ratio of the sampling pipeline length L1 to the sampling pump flow rate V1, i.e., t_sampling = L1 / V1. The transmission pipeline delay time t_transmission is the delay in the transmission of the measurement signal from the sampling point to the CEMS instrument, and is determined by the signal transmission distance L2 and the transmission rate V2, i.e., t_transmission = L2 / V2. The analyzer response time t_response is the time required for the CEMS analyzer to respond to changes in the input signal. The measurement delay time t1 is calculated according to Formula 1. The inlet NOx concentration data collected at the current time t2 is then corrected, that is, the inlet NOx concentration value at time t2 is corrected to the inlet NOx concentration value at time (t2-t1): t1=t_sampling+t_transmission+t_response Formula 1; The calculation module is specifically used for: Let the inlet NOx concentration factor be X(t), and the influencing factors of the inlet NOx concentration be Y(t) = {y1(t), y2(t), y3(t), y4(t), y5(t), y6(t)}, where y1, y2, y3, y4, y5, y6 are the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and pulverizing coal volume, respectively; t is the index number, t = 1, 2, ..., N, and N is the number of inlet NOx concentration samples collected. Then, the Fourier cross-correlation coefficient is determined according to Formula 2: C XY (t)=F^{-1}{X(f)*(Y*(f))} Formula 2; Among them, C XY (t) is the Fourier cross-correlation coefficient of X(t) and Y(t); F^{-1} is the inverse Fourier transform; X(f) and Y(f) are the Fourier transforms of X(t) and Y(t); Y*(f) is the complex conjugate of Y(t); * is the conjugate of a complex number; Calculate the Fourier correlation coefficients of X(t) and Y(t) from time 0 to t3, and calculate the times t4, t5, t6, t7, t8, and t9 corresponding to the maximum values ​​of the Fourier correlation coefficients of y1, y2, y3, y4, y5, and y6. These times are the delay time between the inlet NOx concentration and its influencing factors. Based on the calculated delay times t4, t5, t6, t7, t8, and t9, the data on factors affecting the inlet NOx concentration collected at the current time T are corrected. That is, the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at time T are corrected to the unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal grinding volume at times (T-t4), (T-t5), (T-t6), (T-t7), (T-t8), and (T-t9), respectively. The output variable used for prediction is determined to be the corrected inlet NOx concentration value, and the input variables are the corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume. The prediction model module is specifically used for: The prediction model was constructed using the Zeldovich mechanism model based on factors such as temperature, oxygen concentration, nitrogen concentration, pressure, reaction time, and activation energy. A backpropagation (BP) neural network model is used as the machine learning algorithm. This BP neural network model consists of three layers: an input layer, a hidden layer, and an output layer. The input layer contains six input variables: corrected unit load, total air volume, total coal volume, total primary air volume, total secondary air volume, and coal milling volume. The output layer contains the corrected inlet NOx concentration value. The parameters of the BP neural network model are the weights and thresholds between the input and hidden layers, and between the hidden and output layers. The correction of these weights and thresholds is achieved through a global error function, defined by formulas 3-5. E(θ) = Data(θ) + λPhysics(θ) Equation 3; Where E(θ) is the global error function, Data(θ) is the BP neural network model error function, Physics(θ) is the physical model error function, and λ is the weight coefficient of the physical constraint term, which is obtained empirically or through a genetic optimization algorithm; y i This represents the actual NOx concentration at the inlet. The inlet NOx concentration is the predicted value by the BP neural network model; The inlet NOx concentration is a physical model prediction. By minimizing E(θ), the optimal weights and thresholds between the input layer and the hidden layer, and the optimal weights and thresholds between the hidden layer and the output layer are obtained, thereby determining the prediction model with the optimal physical constraints. Data on factors influencing the inlet NOx concentration at the current moment are collected, and corrections are made based on the calculated delay times t4, t5, t6, t7, t8, and t9. The data are then input into the prediction model with optimal physical constraints to obtain the inlet NOx concentration value at the current moment.

9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method for predicting NOx concentration at the inlet of a thermal power plant denitrification system, taking into account measurement delay, as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an implementation program for information transmission, which, when executed by a processor, implements the steps of the method for predicting the NOx concentration at the inlet of a thermal power plant denitrification system, taking into account measurement delay, as described in any one of claims 1 to 6.

Images