Temperature control method for low-temperature high-NOx-concentration flue gas denitration in nuclear industry

Abstract

The invention discloses a temperature control method for low-temperature high-NOx-concentration flue gas denitration in the nuclear industry, and relates to the technical field of flue gas denitration control. According to the method, flue gas temperature time sequence distribution, NOx concentration gradient and environment radiation intensity data are integrated through a space-time alignment algorithm, a synchronous data matrix representing a system coupling state is generated, then a temperature and efficiency response curved surface is established, a high-efficiency temperature area and an inactivation risk area are dynamically divided through fuzzy logic control, and an optimal temperature regulation and control threshold value is recognized. The drift effect of quantitative radiation on the surface temperature of the catalyst is analyzed through regression, a quantitative relation model of radiation dose and temperature zone deviation is established, and real-time correction of anti-radiation compensation parameters is achieved. And finally, optimizing the multi-objective cost function by adopting a non-dominated sorting genetic algorithm, and obtaining dynamic balance between the maximum NOx conversion rate and the minimum temperature fluctuation. And multi-parameter coupling control of a temperature field and a radiation field is realized, and the denitration efficiency stability under a low-temperature condition is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of flue gas denitration control, and in particular to a temperature control method for denitration of low-temperature and high-NOx concentration flue gas in the nuclear industry. Background Art

[0002] With the increasingly stringent global requirements for environmental protection, nitrogen oxides (NOx), as one of the main atmospheric pollutants, have received widespread attention for their emission control. Especially in the nuclear industry, since low-temperature, high-NOx concentration flue gas may be generated during the operation of nuclear facilities, its denitrification treatment has become an important environmental protection issue. Traditional high-temperature selective catalytic reduction (SCR) technology has been relatively mature in conventional thermal power generation and industrial emission control, but in low-temperature, high-NOx concentration environments, the catalyst activity is reduced and the reaction efficiency is greatly reduced, resulting in existing denitrification technology being difficult to meet the special needs of the nuclear industry. In addition, the radiation factors in the nuclear facility environment may further affect the performance of the catalyst, making the stability of the denitrification system face greater challenges. Therefore, there is an urgent need to develop an efficient temperature control method suitable for low-temperature, high-NOx concentration flue gas in the nuclear industry to improve denitrification efficiency, reduce pollutant emissions, and ensure the long-term reliability and stability of the system.

[0003] Existing low-temperature denitrification methods primarily include low-temperature SCR catalyst modification, plasma-assisted denitrification, and oxidative adsorption. However, these methods still face numerous challenges in practical application. First, while low-temperature SCR catalysts can maintain a certain level of activity at relatively low temperatures, they suffer from poor durability and are susceptible to sulfur poisoning and carbon deposition, leading to rapid degradation of catalyst activity. Second, while plasma-assisted denitrification demonstrates high denitrification efficiency in laboratory settings, it consumes a high amount of energy, and the plasma action zone struggles to uniformly cover the entire flue gas flow field, impacting reaction efficiency. Furthermore, oxidative adsorption suffers from a low reaction rate at low temperatures and limited adsorbent regeneration, resulting in high long-term operating costs. Furthermore, existing temperature control methods typically rely on fixed temperature set points or simple PID control, failing to account for the effects of NOx concentration gradients, flue gas temperature fluctuations, and ambient radiation on catalyst performance. This results in insufficient temperature control precision and makes dynamic optimization difficult. Therefore, to meet the specific needs of low-temperature, high-NOx flue gas denitrification in the nuclear industry, an intelligent temperature control method is urgently needed that integrates dynamic NOx concentration changes, temperature response characteristics, and radiation effects to achieve more efficient and stable denitrification. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry, which solves the problems of the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry, comprising the following steps: S1. real-time acquisition of flue gas temperature time series distribution, NOx concentration gradient change, and ambient radiation intensity data, eliminating sensor acquisition delays through a spatiotemporal alignment algorithm, generating a synchronized data matrix, and characterizing the coupled state of temperature, concentration, and radiation in the reactor; S2. analyzing the nonlinear effect of temperature fluctuations on NOx conversion rate based on the synchronized data matrix, identifying the catalyst activity sensitive range, establishing a temperature and efficiency response surface, dividing the high-efficiency temperature zone into the deactivation risk temperature zone, and dynamically outputting temperature control thresholds and priority weights through a fuzzy logic control algorithm; S3. associating the response surface with the radiation intensity data, quantifying the drift effect of radiation on the catalyst surface temperature, fitting the functional relationship between radiation dose and temperature zone offset through regression analysis, correcting the temperature control threshold, and generating an anti-radiation compensation parameter; S4. constructing a multi-objective cost function based on the corrected temperature control threshold and anti-radiation compensation parameter with the goals of maximizing NOx conversion efficiency and minimizing temperature fluctuations, optimizing the multi-objective cost function through a non-dominated sorting genetic algorithm, and outputting an optimal temperature control instruction.

[0006] Furthermore, the time series distribution of flue gas temperature, NOx concentration gradient change and ambient radiation intensity data are obtained in real time, and the sensor acquisition delay is eliminated through the time-space alignment algorithm. The specific process of generating a synchronous data matrix is ​​as follows: according to the sensor layout position and flue gas flow rate information, the signal propagation time difference of each measurement point is calculated, and the sensor clock drift is dynamically corrected through the Kalman filter algorithm; the collected time series signals of flue gas temperature, NOx concentration and radiation intensity are aligned through the dynamic time warping algorithm; the covariance matrix of each signal is analyzed through covariance, abnormal data is detected and eliminated, and a synchronous data matrix is ​​generated.

[0007] Furthermore, based on the synchronized data matrix, the nonlinear effect of temperature fluctuations on NOx conversion rate is analyzed, and the specific process of identifying the sensitive range of catalyst activity is as follows: based on the synchronized data matrix, the temperature and NOx concentration are mapped to the high-dimensional feature space through the Gaussian kernel function, the principal component contribution rate is calculated, the temperature and concentration correlation characteristics are extracted, and the linear and nonlinear response components are separated; based on the dynamic sliding window statistics, the cumulative effect of temperature fluctuations on the conversion rate is calculated, different window lengths are set, and the time response curve of temperature fluctuations to NOx conversion rate is calculated to identify the critical temperature points that cause catalyst deactivation or efficient reaction; the temperature and conversion rate distributions are clustered and analyzed by the clustering algorithm to divide the sensitive range of catalyst activity.

[0008] Furthermore, the specific process of establishing temperature and efficiency response surfaces and dividing high-efficiency temperature zones and deactivation risk temperature zones is as follows: by synchronizing the temperature and NOx conversion rate data in the data matrix, a response surface of temperature and NOx conversion efficiency is established through regression analysis method; on the response surface, the influence of temperature change on NOx conversion rate is identified, and conversion efficiency functions under different temperature ranges are established; according to the conversion efficiency values ​​on the response surface, temperature zones with higher NOx conversion rates and deactivation risk temperature zones with lower conversion efficiency are divided; by analyzing the catalyst activity under different temperature ranges, the optimal reaction temperature range and the risk temperature range of catalyst deactivation are determined.

[0009] Furthermore, the specific process of dynamically outputting the temperature control threshold and priority weight through the fuzzy logic control algorithm is as follows: based on the established high-efficiency temperature zone and deactivation risk temperature zone response surface, the temperature control threshold and priority weight are dynamically output through the fuzzy logic control algorithm, including: setting the temperature fluctuation amplitude and catalyst activity state in the reactor as input fuzzy variables, and setting the temperature control threshold and priority weight as output fuzzy variables; setting fuzzy rules to define the fuzzy rules between the temperature fluctuation amplitude, catalyst activity and temperature control threshold and priority weight; through the fuzzy control rule base, the fuzzified input variables are inferred, and the membership of each fuzzy set is comprehensively considered to obtain the corresponding temperature control threshold and priority weight.

[0010] Furthermore, the response surface is associated with the radiation intensity data to quantify the drift effect of radiation on the catalyst surface temperature. The specific process is as follows: extract the temperature zone offset characteristics in the response surface and construct a correlation matrix with the radiation intensity time series data; analyze the time-frequency correlation between radiation and temperature drift through cross-wavelet transform; quantify the contribution of radiation dose to the high-efficiency temperature zone boundary offset and generate a drift coefficient matrix.

[0011] Furthermore, the functional relationship between radiation dose and temperature zone offset is fitted through regression analysis, the temperature control threshold is corrected, and the specific process of generating anti-radiation compensation parameters is as follows: the nonlinear relationship between radiation dose and temperature zone offset is fitted by using a polynomial kernel regression model, and high-order nonlinear terms are retained; in order to avoid overfitting of the regression model, radiation-related factors that have a significant impact on the temperature zone offset are screened out through elastic network regression; the regression coefficient is embedded in the threshold correction equation, and the anti-radiation compensation parameters are output.

[0012] Furthermore, based on the modified temperature control threshold and anti-radiation compensation parameters, with the goal of maximizing NOx conversion efficiency and minimizing temperature fluctuations, the specific process of constructing a multi-objective cost function is as follows: based on the modified temperature threshold, the response surface integral area is constrained to maximize NOx conversion efficiency; the temperature time series variance and radiation drift residual are constrained to minimize temperature fluctuations; the dual objectives are fused through the dynamic weight allocation function to generate a multi-objective cost function.

[0013] Furthermore, the logic of optimizing the multi-objective cost function through the non-dominated sorting genetic algorithm is as follows: initialize the population: use the temperature control instruction as the genetic code to randomly generate a set of candidate solutions; fast non-dominated sorting: calculate the cost function value of each individual in the population, sort the solutions according to the cost function value, and divide the Pareto front level according to the cost function value; crowding calculation and elite retention: when performing the selection operation, calculate the crowding of the solution set and retain the elite solution to ensure the diversity and convergence of the solution set, and iteratively output the optimal temperature control instruction.

[0014] The present invention has the following beneficial effects:

[0015] (1) A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry eliminates sensor data acquisition delays through a spatiotemporal alignment algorithm, ensures the synchronization of flue gas temperature, NOx concentration gradient, and ambient radiation intensity data, and improves the accuracy of reactor state characterization; based on a fuzzy logic control algorithm, it dynamically identifies the catalyst activity sensitive range, establishes a temperature-efficiency response surface, and adaptively adjusts the temperature control threshold and priority weight according to the temperature zone characteristics, thereby improving the NOx conversion rate, avoiding premature catalyst deactivation, and enhancing the stability and reliability of the system.

[0016] (2) A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry is developed. The method quantifies the drift effect of radiation on the catalyst surface temperature, fits the relationship between radiation dose and temperature zone offset through regression analysis, corrects the temperature control threshold and generates anti-radiation compensation parameters, and improves the adaptability of the temperature control strategy in the nuclear industry environment. Based on the multi-objective optimization method, an optimization cost function is constructed with the goals of maximizing NOx conversion efficiency and minimizing temperature fluctuations. The optimal temperature control instruction is solved through a non-dominated sorting genetic algorithm to achieve refined temperature control, improve denitrification efficiency and reduce system energy consumption.

[0017] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to the present invention. DETAILED DESCRIPTION

[0019] This application demonstrates a temperature control method for denitrification of low-temperature, high-NOx-concentration flue gas in the nuclear industry, addressing the issues of reduced NOx conversion and increased risk of catalyst deactivation caused by reactor temperature fluctuations. This method utilizes a spatiotemporal alignment algorithm to improve data synchronization, employs a fuzzy logic control algorithm to dynamically adjust temperature control thresholds, and combines radiation impact modeling with a multi-objective optimization strategy to optimize temperature control accuracy, thereby improving NOx removal efficiency and extending catalyst life.

[0020] The overall idea of ​​the solution in the embodiments of this application is as follows:

[0021] The flue gas temperature time series distribution, NOx concentration gradient change and ambient radiation intensity data are acquired in real time. The sensor acquisition delay is eliminated through the time-space alignment algorithm to generate a synchronous data matrix to characterize the coupling state of temperature, concentration and radiation in the reactor.

[0022] Based on the synchronized data matrix, the nonlinear effect of temperature fluctuations on NOx conversion rate is analyzed, the sensitive range of catalyst activity is identified, the temperature and efficiency response surface is established, the high-efficiency temperature zone and the deactivation risk temperature zone are divided, and the temperature control threshold and priority weight are dynamically output through the fuzzy logic control algorithm.

[0023] The response surface is associated with the radiation intensity data to quantify the drift effect of radiation on the catalyst surface temperature. The functional relationship between the radiation dose and the temperature zone offset is fitted through regression analysis, the temperature control threshold is corrected, and the anti-radiation compensation parameter is generated.

[0024] According to the modified temperature control threshold and anti-radiation compensation parameters, a multi-objective cost function is constructed with the goal of maximizing NOx conversion efficiency and minimizing temperature fluctuations. The multi-objective cost function is optimized through a non-dominated sorting genetic algorithm to output the optimal temperature control instruction.

[0025] See also Figure 1An embodiment of the present invention provides a technical solution: a temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry, comprising the following steps: S1. acquiring real-time flue gas temperature time series distribution, NOx concentration gradient change, and ambient radiation intensity data, eliminating sensor acquisition delays through a spatiotemporal alignment algorithm, generating a synchronized data matrix, and characterizing the coupled state of temperature, concentration, and radiation in the reactor; S2. analyzing the nonlinear effect of temperature fluctuations on NOx conversion rate based on the synchronized data matrix, identifying the catalyst activity sensitive range, establishing a temperature and efficiency response surface, dividing the high-efficiency temperature zone into a deactivation risk temperature zone, and dynamically outputting a temperature control threshold and priority weight through a fuzzy logic control algorithm; S3. correlating the response surface with the radiation intensity data, quantifying the drift effect of radiation on the catalyst surface temperature, fitting the functional relationship between radiation dose and temperature zone offset through regression analysis, correcting the temperature control threshold, and generating an anti-radiation compensation parameter; S4. constructing a multi-objective cost function based on the corrected temperature control threshold and anti-radiation compensation parameter with the goals of maximizing NOx conversion efficiency and minimizing temperature fluctuations, optimizing the multi-objective cost function through a non-dominated sorting genetic algorithm, and outputting an optimal temperature control instruction.

[0026] In this implementation scheme, S1: Due to the differences in the spatiotemporal distribution of the reactor's internal temperature, NOx concentration, and ambient radiation, there may be delays or mismatches in the different data sources collected by the sensors. This step uses a spatiotemporal alignment algorithm to synchronize the temperature, concentration, and radiation intensity data, eliminate acquisition deviations, and form a highly timely and high-precision synchronized data matrix to provide reliable data support for subsequent modeling. Spatiotemporal alignment algorithm: This algorithm is used to eliminate the time lag and spatial offset of sensor data, ensuring that data such as temperature, NOx concentration, and radiation intensity are aligned on the same time basis to form a synchronized data matrix, allowing the control system to make judgments based on accurate data. Synchronous data matrix: refers to a matrix data structure that uniformly characterizes temperature, concentration, and radiation on the same time scale, and can accurately reflect the temperature-concentration-radiation coupling state in the reactor. S2 (Analysis and fuzzy control of the impact of temperature fluctuations on NOx conversion rate): The reactor temperature has a nonlinear effect on the NOx conversion rate. In certain temperature ranges, the catalyst activity is more sensitive, and slight fluctuations may affect the denitrification efficiency or even accelerate catalyst deactivation. This step establishes a temperature-efficiency response surface based on experimental or simulation data, divides the high-efficiency temperature zone and the deactivation risk temperature zone, and uses a fuzzy logic control algorithm to dynamically calculate the temperature control threshold and priority weight to achieve flexible and accurate temperature control. Nonlinear effect: refers to the effect of temperature on NOx conversion rate is not a simple linear relationship, but a complex curve change. In certain specific temperature ranges, the catalyst activity is optimal, while in other ranges it shows attenuation or deactivation. Temperature-efficiency response surface: A three-dimensional surface model drawn through experimental or simulation data, which is used to describe the changing trend of NOx conversion efficiency at different temperatures, thereby identifying high-efficiency temperature zones and deactivation risk temperature zones. Fuzzy logic control algorithm: This algorithm is suitable for nonlinear and uncertain systems. By setting temperature control thresholds and priority weights, it can achieve dynamic adjustment of temperature so that it is always in the high-efficiency temperature zone and avoids entering the deactivation temperature zone. S3 (radiation effect compensation): In a nuclear industry environment, radiation may cause catalyst surface temperature drift, affecting the stability of the denitrification reaction. This step associates the temperature-efficiency response surface with the ambient radiation intensity data, analyzes the time-frequency relationship between radiation and temperature drift through cross-wavelet transform, and uses regression analysis to fit the functional relationship between radiation dose and temperature zone offset. Finally, the temperature control threshold is corrected to generate anti-radiation compensation parameters to improve the adaptability of the temperature control system in the radiation environment. Radiation dose: refers to the nuclear radiation energy received by the catalyst surface per unit time, which has a direct impact on the catalyst temperature. Temperature zone offset: the change in the optimal temperature zone position due to radiation, that is, the drift of the optimal activity temperature of the catalyst. Cross-wavelet transform: a signal analysis method used to analyze the correlation between different physical quantities (such as radiation intensity and catalyst temperature) in the time-frequency domain, so as to quantify the impact of radiation on temperature.Regression analysis: Through mathematical modeling methods, the functional relationship between radiation dose and temperature zone offset is fitted to generate anti-radiation compensation parameters and modify the temperature control strategy. S4 (multi-objective optimization and optimal temperature control instruction output): While ensuring the NOx conversion rate, the temperature fluctuation should be as small as possible to improve the catalyst life and system stability. This step constructs a multi-objective cost function, where: Goal 1: Maximize NOx conversion efficiency, based on the area of ​​the response curve integral constrained by the corrected temperature threshold; Goal 2: Minimize temperature fluctuation, limit the temperature time series variance and radiation drift residual. The cost function is optimized using the non-dominated sorting genetic algorithm (NSGA-II) to obtain the optimal temperature control instruction, which is fed back to the system to achieve closed-loop optimization control. Multi-objective cost function: Set two optimization goals: Goal 1: Maximize NOx conversion efficiency, that is, keep the temperature in the optimal catalytic range; Goal 2: Minimize temperature fluctuation, that is, control the temperature change amplitude of the system as small as possible to protect the catalyst life. Non-dominated sorting genetic algorithm: An optimization algorithm suitable for optimization problems with multiple conflicting objectives. It simulates natural selection and evolution mechanisms to find a set of optimal solutions, ultimately outputting optimal temperature control instructions and feeding them back to the system to achieve automatic temperature regulation.

[0027] Specifically, the time series distribution of flue gas temperature, NOx concentration gradient change and ambient radiation intensity data are acquired in real time, and the sensor acquisition delay is eliminated through the space-time alignment algorithm. The specific process of generating a synchronous data matrix is ​​as follows: according to the sensor layout position and flue gas flow rate information, the signal propagation time difference of each measurement point is calculated, and the sensor clock drift is dynamically corrected through the Kalman filter algorithm; the collected time series signals of flue gas temperature, NOx concentration and radiation intensity are aligned through the dynamic time warping algorithm; the covariance matrix of each signal is analyzed through covariance, abnormal data is detected and eliminated, and a synchronous data matrix is ​​generated.

[0028] In this embodiment, the signal propagation time difference is calculated. Since the sensors are distributed in different positions, there is a time difference in data collection. According to the position of the sensor and the flue gas flow rate, the signal propagation time difference is calculated to ensure that the data can be analyzed under the same time reference, thereby improving the control accuracy. Correct the sensor clock drift. The clocks of different sensors may drift, resulting in inconsistent data timestamps. The Kalman filter algorithm is used to dynamically correct the clock error of the sensor so that the data of all sensors can be synchronized under the same time reference to avoid time deviation affecting the analysis results. Timing signal alignment. Sensor data may be collected at different time points, so the timing signal needs to be aligned. The dynamic time warping (DTW) algorithm is used to match sensor data at different times so that all sensor data can be compared at the same time point, thereby improving the consistency and availability of the data. Abnormal data removal. During the data collection process, the sensor may be subject to noise or interference, resulting in abnormal data. Covariance analysis is used to detect and remove these abnormal data points to ensure that the final data matrix is ​​not interfered with and ensure the accuracy of subsequent analysis. Covariance analysis: This method is used to calculate the correlation of sensor data. By analyzing the covariance matrix of different signals, abnormal data points are detected and outliers are removed. Covariance matrix: in: is the variance of flue gas temperature; is the variance of NOx concentration; is the variance of radiation intensity; σ T,Nox Represents the covariance between temperature and NOx concentration; other items are similar. If the data of a certain measurement point deviates too much from the overall trend (that is, exceeds the set standard deviation threshold), the data point is considered abnormal and is removed. After time correction, data alignment, and anomaly removal, a synchronized data matrix is ​​finally generated for subsequent temperature control calculations. Synchronized data matrix: This matrix records the flue gas temperature, NOx concentration, and ambient radiation intensity under the same time reference. An example is as follows:

[0029]

[0030] This matrix is ​​used to analyze the impact of temperature fluctuations on NOx conversion rate and the offset effect of radiation on catalyst surface temperature, providing data support for subsequent temperature control strategies.

[0031] Specifically, based on the synchronized data matrix, the nonlinear effect of temperature fluctuations on NOx conversion rate is analyzed, and the specific process of identifying the sensitive range of catalyst activity is as follows: based on the synchronized data matrix, the temperature and NOx concentration are mapped to the high-dimensional feature space through the Gaussian kernel function, the principal component contribution rate is calculated, the temperature and concentration correlation characteristics are extracted, and the linear and nonlinear response components are separated; based on the dynamic sliding window statistics, the cumulative effect of temperature fluctuations on the conversion rate is calculated, different window lengths are set, and the time response curve of temperature fluctuations to NOx conversion rate is calculated to identify the critical temperature points that cause catalyst deactivation or efficient reaction; the temperature and conversion rate distributions are clustered and analyzed by the clustering algorithm to divide the sensitive range of catalyst activity.

[0032] In this embodiment, in order to solve the nonlinear relationship between temperature and NOx concentration, a Gaussian kernel function is used to map the temperature and concentration data into a high-dimensional feature space. The form of the Gaussian kernel function is: Among them, x and y represent two data points of temperature and NOx concentration respectively, and σ is the bandwidth parameter of the kernel function, which controls the similarity between data points. Through this mapping, the nonlinear relationship between temperature and NOx concentration can be better represented in high-dimensional space. Calculate the principal component contribution rate, extract the temperature and concentration correlation features, and use principal component analysis (PCA) to extract the correlation features of temperature and NOx concentration. The goal of principal component analysis is to transform the data into a new coordinate system so that the first principal component has the largest variance, and the second principal component is orthogonal to the first principal component and has the second largest variance. Calculate the contribution rate of each principal component, the formula is: Among them, WER is the contribution rate, λ a is the eigenvalue of the ath principal component, which represents the variance of the principal component, and A is the total number of eigenvalues. In this way, the most important features can be extracted from high-dimensional data to help identify the correlation between temperature fluctuations and NOx concentrations. By calculating the contribution rate of the principal component, it is possible to identify which parts of the response of temperature fluctuations to NOx conversion rate are linear and which are nonlinear. Usually, linear responses can be represented by low-order features (such as the first few components of the principal component), while nonlinear responses may depend on higher-order features or nonlinear models. The cumulative effect of temperature fluctuations on conversion rate based on dynamic sliding window statistics: In order to analyze the dynamic impact of temperature fluctuations on NOx conversion rate, a dynamic sliding window method is used. The data in the window is used to calculate the impact of temperature fluctuations on conversion rate. The window length is set to L, and the changes in temperature and NOx concentration in the window are gradually calculated through the sliding window to calculate the impact of each time period. The calculation formula is: Where, ΔT avg (t) is the average temperature fluctuation in the window at time t, T i is the temperature value at the i-th moment, T avgIt is the average value of the temperature in the window. According to the fluctuation value in the window, the influence of temperature fluctuation on NOx conversion rate is analyzed. Calculate the time response curve of temperature fluctuation on NOx conversion rate: The time response curve of temperature fluctuation can be compared with the time series data of NOx conversion rate through the results of sliding window. Specifically, the influence of temperature fluctuation on NOx conversion rate can be established by regression analysis to establish a time response model, and there is a linear or nonlinear relationship between temperature fluctuation and NOx conversion rate. Through the regression model, the time response curve of temperature fluctuation on conversion rate is obtained. C(t)=β0+β1ΔT avg (t)+β2ΔT avg (t) 2 +β3ΔT avg (t) 3 +…+β n ΔT avg (t) n ; Where: C(t) is the NOx conversion rate at time t. ΔT avg (t) is the average value of the temperature fluctuation calculated at time t (using a sliding window approach). β0 is the constant term or intercept of the regression model, representing the baseline conversion rate when the temperature fluctuation is zero. β1,β2,…,β n are regression coefficients, representing the effects of different powers of temperature fluctuation on conversion. n is the highest-order term in the regression model. Based on the response curve of temperature fluctuation to conversion, the critical point of NOx conversion change is identified. This critical point corresponds to the turning point in catalyst activity, marking catalyst deactivation or entry into a high-efficiency reaction range. Specifically, through peak detection or inflection point analysis, the temperature value at which the conversion rate fluctuates most dramatically is identified as the critical temperature point. Cluster analysis of temperature and conversion rate distributions is performed using a clustering algorithm: To identify catalyst activity-sensitive ranges, a clustering algorithm can be used to cluster the temperature and NOx conversion rate distributions. The clustering algorithm divides the temperature and conversion rate data into multiple clusters, each representing a different relationship between temperature and conversion rate. The clustering results can help distinguish the catalyst's high-efficiency reaction zone from its deactivation risk zone, further delineating the catalyst's activity-sensitive range. These steps, combined with high-dimensional feature extraction, sliding window analysis, and clustering algorithms, can effectively identify the catalyst's sensitive range, providing an important reference for temperature control.

[0033] Specifically, the specific process of establishing temperature and efficiency response surfaces and dividing high-efficiency temperature zones and deactivation risk temperature zones is as follows: by synchronizing the temperature and NOx conversion rate data in the data matrix, a response surface of temperature and NOx conversion efficiency is established through regression analysis method; on the response surface, the impact of temperature changes on NOx conversion rate is identified, and conversion efficiency functions under different temperature ranges are established; according to the conversion efficiency values ​​on the response surface, temperature zones with higher NOx conversion rates and deactivation risk temperature zones with lower conversion efficiency are divided; by analyzing the catalyst activity under different temperature ranges, the optimal reaction temperature range and the risk temperature range of catalyst deactivation are determined.

[0034] In this embodiment, a response surface of temperature and NOx conversion efficiency is established by regression analysis: by synchronizing the temperature and NOx conversion rate data in the data matrix, a regression analysis method (such as linear regression or polynomial regression) is used to establish a response surface between temperature and NOx conversion efficiency. This process finds the functional relationship between temperature and NOx conversion efficiency by fitting the data points, and obtains the response relationship between temperature and conversion efficiency. The effect of temperature change on NOx conversion rate is identified. After obtaining the response surface, the effect of temperature change on NOx conversion rate can be identified by analyzing the slope or rate of change of the surface. For example, the sensitivity of temperature change to conversion rate can be calculated: Where S(T) represents the sensitivity of temperature to conversion rate, E(T) is the conversion efficiency, and T is temperature. By calculating the sensitivity at different temperatures, we can find the range where temperature fluctuations significantly affect NOx conversion rate. We can establish conversion efficiency functions for different temperature ranges. Based on the regression analysis results, we can establish conversion efficiency functions for different temperature ranges. For each temperature range, we can calculate the conversion efficiency corresponding to the temperature. For example, we can set several temperature ranges T1, T2, ..., Tk, and then calculate the average conversion efficiency in each range: Among them, E avg (T k ) is the average conversion efficiency of the kth temperature interval, T is the temperature in the interval, N k Is the number of temperature points in this range. Divide the temperature zone with higher NOx conversion rate and the deactivation risk zone with lower conversion efficiency. By analyzing the conversion efficiency in different temperature ranges, high efficiency zone and deactivation risk zone can be divided. Specifically, set the conversion efficiency threshold E thresh , when the conversion efficiency E(T k ) is greater than the threshold, the corresponding temperature zone is the high efficiency zone; when the conversion efficiency is less than the threshold, the corresponding temperature zone is the inactivation risk zone. That is: high efficiency zone: E(T k )>E thresh ; Inactivation risk area: E(T k )≤E threshAfter determining the conversion efficiency in different temperature ranges, further analyze the activity and deactivation risk of the catalyst based on the conversion efficiency. Generally, the high conversion efficiency range corresponds to the optimal reaction temperature range of the catalyst, while the low conversion efficiency range corresponds to the deactivation risk temperature range of the catalyst. The maximum efficiency point can be found based on the conversion efficiency curve, and the optimal reaction temperature range can be set around it: T opt =argmaxE(T); where T opt represents the optimal reaction temperature. The inactivation risk temperature range is the temperature range where the conversion efficiency decreases significantly. The corresponding temperature range can be determined by analyzing the inflection point or the efficiency drop amplitude of the response surface.

[0035] Specifically, the specific process of dynamically outputting temperature control thresholds and priority weights through the fuzzy logic control algorithm is as follows: based on the established high-efficiency temperature zone and deactivation risk temperature zone response surface, the temperature control thresholds and priority weights are dynamically output through the fuzzy logic control algorithm, including: setting the temperature fluctuation amplitude and catalyst activity state in the reactor as input fuzzy variables, and setting the temperature control thresholds and priority weights as output fuzzy variables; setting fuzzy rules to define the fuzzy rules between the temperature fluctuation amplitude, catalyst activity and temperature control thresholds and priority weights; through the fuzzy control rule base, reasoning on the fuzzified input variables, comprehensively considering the membership of each fuzzy set, and obtaining the corresponding temperature control thresholds and priority weights.

[0036] In this implementation, input and output fuzzy variables are defined as follows: Input fuzzy variables: Temperature fluctuation amplitude: The range of temperature fluctuation within the reactor, representing the severity of temperature changes. Generally speaking, large temperature fluctuations may affect the catalyst's reaction efficiency, requiring regulation to reduce the fluctuations. Catalyst activity: The catalyst's reaction efficiency is characterized by temperature and NOx concentration. High catalyst activity indicates good reaction efficiency; low activity may require temperature adjustment to improve reaction efficiency. Output fuzzy variables: Temperature control threshold: In the reactor, a temperature control threshold is dynamically output based on the temperature response surface, used to determine whether temperature adjustment is necessary. Priority weight: Used to indicate the priority level to consider when adjusting temperature. For example, when catalyst activity is low, increasing the temperature may be a priority. Fuzzy rule setting: Fuzzy rules define the relationship between input and output fuzzy variables. The fuzzy rules are as follows: Rule 1: If the temperature fluctuation is large and the catalyst activity is low, the temperature control threshold is low (i.e., the temperature fluctuation requires significant adjustment) and the priority weight is high (prioritizing temperature adjustment). Rule 2: If the temperature fluctuation is small and the catalyst activity is high, the temperature control threshold is higher (i.e., the temperature change is small and does not require significant adjustment) and the priority weight is lower (i.e., the temperature change has a lower priority). These rules are based on the reactor's temperature response surface and the catalyst's activity sensitivity range, and are set based on experience and actual needs. In the fuzzy logic control algorithm, the input variables (such as temperature fluctuation and catalyst activity) are first fuzzified. This means converting these input variables from precise numerical values ​​into membership functions that represent their membership in various fuzzy sets. Membership functions describe the membership relationships between input variables and fuzzy sets. For example, a membership function for temperature fluctuation might have the fuzzy sets "low," "medium," and "high." The fuzzy inference process involves applying fuzzy reasoning methods to a fuzzy rule base, combining the fuzzified input variables with fuzzy rules to derive fuzzy values ​​for the output fuzzy variables (temperature control threshold and priority weight). For example, if the temperature fluctuation is large and the catalyst activity is low, the output temperature control threshold might be "low" and the priority weight might be "high." If the temperature fluctuation is small and the catalyst activity is high, the output temperature control threshold may be "higher" and the priority weight is "lower". Defuzzification: Finally, the result of fuzzy reasoning is converted into precise numerical values ​​through the defuzzification process (usually using the centroid method or maximum membership method) to obtain the temperature control threshold and priority weight. The precise output value is obtained by calculating the "centroid" of the output membership function. For example, if the output variable is "temperature control threshold", the result obtained after defuzzification is a specific numerical value. Output temperature control threshold and priority weight, the system will output two precise numerical values ​​based on the reasoning results: Temperature control threshold: used to adjust the temperature in the reactor.Priority Weight: Used to determine which factor has higher priority in the temperature control process.

[0037] Specifically, the specific process of associating the response surface with the radiation intensity data and quantifying the drift effect of radiation on the catalyst surface temperature is as follows: extracting the temperature zone offset characteristics in the response surface and constructing a correlation matrix with the radiation intensity time series data; analyzing the time-frequency correlation between radiation and temperature drift through cross-wavelet transform; quantifying the contribution of radiation dose to the high-efficiency temperature zone boundary offset and generating a drift coefficient matrix.

[0038] In this embodiment, the response surface is a graph of the relationship between temperature and NOx conversion efficiency, which shows the conversion efficiency under different temperature ranges. In order to quantify the effect of radiation on the catalyst temperature, it is first necessary to extract the temperature zone offset features from the response surface. These offset features represent the offset of the boundary of the temperature response surface under different radiation intensity conditions. The method for extracting the offset features can determine the impact of radiation intensity on the temperature distribution by analyzing the temperature values ​​in different temperature ranges. For example, if the radiation intensity is high, it may cause the upper limit (or boundary) of the efficient temperature zone in the reactor to shift to a higher temperature. In order to associate the temperature zone offset features with the radiation intensity data, we compare the temperature range features on the temperature response surface with the time series data of the radiation intensity to construct a correlation matrix. The rows of the matrix represent different temperature ranges or temperature offset features, and the columns represent the radiation intensity values ​​at different time points. Through this matrix, we can observe the relationship between radiation intensity and temperature zone offset at each time point. Cross wavelet transform is a time-frequency analysis method used to analyze the correlation between two signals at different frequencies. Through cross wavelet transform, the time-frequency correlation between radiation intensity and temperature drift can be revealed, that is, how radiation affects temperature fluctuations at different time periods and frequencies. In this step, we perform cross-wavelet transform on the radiation intensity data and temperature drift data (through the extracted temperature zone offset features). The core goal of the cross-wavelet transform is to analyze the mutual influence of the two signals in the time-frequency domain and reveal the phase difference and correlation between them. After completing the cross-wavelet transform, we can quantify the impact of radiation dose on the boundary of the high-efficiency temperature zone. The purpose of this step is to calculate the contribution of radiation intensity to temperature drift, especially in the boundary part of the high-efficiency temperature zone. By analyzing the results of the cross-wavelet transform and combining the temperature zone offset features, we can obtain the specific impact of radiation dose on the boundary of the high-efficiency temperature zone, which is expressed as a "drift coefficient". The drift coefficient can be calculated by fitting the relationship between the temperature offset and radiation intensity in the cross-wavelet transform results by the least squares method: and the relationship between the radiation intensity: Where: ΔT z Indicates the temperature offset value at the zth moment, R zis the corresponding radiation intensity value, and α is the drift coefficient, which indicates the degree of influence of radiation intensity on the temperature zone boundary. By calculating the drift coefficient α, we can quantify the degree of influence of radiation dose on the temperature zone boundary, and further analyze how the temperature zone boundary shifts under different radiation conditions. Generate a drift coefficient matrix. By quantifying the contribution of radiation to temperature drift at each time point (i.e., drift coefficient), we generate a drift coefficient matrix that represents the degree of temperature zone boundary drift at different time points and different radiation intensities. The elements of this matrix can be expressed as: D = (α1α2…α n ); where α i Indicates the contribution of radiation intensity to temperature zone deviation at moment i.

[0039] Specifically, the functional relationship between radiation dose and temperature zone offset is fitted through regression analysis, the temperature control threshold is corrected, and the specific process of generating anti-radiation compensation parameters is as follows: the nonlinear relationship between radiation dose and temperature zone offset is fitted by using a polynomial kernel regression model, and high-order nonlinear terms are retained; in order to avoid overfitting of the regression model, radiation-related factors that have a significant impact on the temperature zone offset are screened out through elastic network regression; the regression coefficient is embedded in the threshold correction equation, and the anti-radiation compensation parameters are output.

[0040] In this embodiment, there may be a nonlinear relationship between the radiation dose and the temperature zone offset. To this end, we use a polynomial kernel regression model for fitting to capture this nonlinear relationship. The core idea of ​​polynomial kernel regression is to map the data into a higher-dimensional feature space through kernel techniques and perform regression analysis in this space. In polynomial kernel regression, the input variable is set to the radiation dose R and the output variable is set to the temperature zone offset ΔT. The mathematical expression of the regression model is: Where: R i is the radiation dose in the training data, K(R i ,R) is the radiation dose R i The kernel function between and R is usually a polynomial kernel function: K(R i ,R)=(R i +c) d ; where c is a constant and d is the order of the polynomial, controlling the degree of nonlinearity. α iis the weight of each training sample, and b is the bias term. Through the kernel regression model, we can obtain the nonlinear fitting relationship between radiation dose and temperature zone offset. Elastic network regression screens significant radiation-related factors. In the polynomial regression model, multiple radiation-related factors may be included, but not all factors have a significant impact on temperature zone offset. In order to avoid overfitting and screen out factors that have an important impact on temperature zone offset, we use the elastic network regression method. Elastic network regression is a combination of L1 regularization (Lasso) and L2 regularization (Ridge), which can handle multicollinearity and effectively select features. Its loss function is: Where: ΔT i is the temperature zone offset at moment i, R ij is the radiation factor at moment i, β j is the regression coefficient, which indicates the degree of influence of the radiation factor. λ1 and λ2 are regularization parameters, which control the penalty strength of Lasso and Ridge respectively. Through elastic network regression, we can screen out radiation-related factors that have a significant impact on temperature zone offset, ensuring the simplicity and robustness of the regression model. After regression analysis and feature screening, we can obtain the relationship coefficient between the radiation factor and the temperature zone offset, that is, the regression coefficient R. j These coefficients reflect the degree of influence of each radiation factor on the temperature zone deviation. Next, we embed the regression coefficients into the temperature control threshold correction equation. Set the initial temperature control threshold to Tth, and the corrected temperature threshold T lh ' can be expressed as: Where: T th is the initial temperature control threshold; β j is the regression coefficient of the radiation factor; R j is the jth radiation factor. Using this formula, the temperature control threshold is dynamically modified based on the influence of varying radiation intensities, thereby optimizing the temperature control strategy. This generates an anti-radiation compensation parameter, which, along with the corresponding regression coefficient, constitutes the anti-radiation compensation parameter. This parameter is used to automatically adjust the temperature control strategy based on changes in radiation intensity during the actual reaction process, ensuring that the catalyst maintains efficient operation under the influence of radiation. The mathematical representation of the anti-radiation compensation parameter γ is: Where: j is the regression coefficient of the radiation factor, R j is the jth radiation factor.

[0041] Specifically, according to the modified temperature control threshold and anti-radiation compensation parameters, with the goal of maximizing NOx conversion efficiency and minimizing temperature fluctuations, the specific process of constructing a multi-objective cost function is as follows: based on the modified temperature threshold, the response surface integral area is constrained to maximize NOx conversion efficiency; the temperature time series variance and radiation drift residual are constrained to minimize temperature fluctuations; the dual objectives are fused through the dynamic weight allocation function to generate a multi-objective cost function.

[0042] In this embodiment, in order to maximize the NOx conversion efficiency, we use the modified temperature control threshold to constrain the integral area of ​​the response curve, which represents the NOx conversion efficiency within a certain temperature range. Specifically, the total integral of the conversion efficiency can be expressed as: Where: J1 represents the NOx conversion efficiency objective function, C(T) is the NOx conversion rate at temperature T, T min and T max are the minimum and maximum values ​​of the temperature. The goal of this section is to optimize the efficiency of the catalytic reaction by maximizing the NOx conversion rate. The control of temperature fluctuations is accomplished by minimizing the temperature time series variance and the radiation drift error. The temperature fluctuation (variance) and radiation drift error can be calculated as follows: in: is the variance of temperature, indicating temperature fluctuation; T(t) is the temperature data at time t; is the average temperature. Radiation drift error ∈ rad is also included in the measure of temperature fluctuation, which is expressed as: ∈ rad =T(t)-T rad (t); where: T rad (t) is the temperature predicted by the radiation model, and T(t) is the actual measured temperature. The goal of temperature fluctuation is to maximize the above variance and radiation drift error, so as to stabilize temperature fluctuations and ensure the stable operation of the catalyst. Constructing a multi-objective cost function Based on the above two objectives, we combine the maximization of NOx conversion efficiency and the minimization of temperature fluctuations into a comprehensive multi-objective cost function. In this process, we need to use a dynamic weight allocation function to adjust the priority of each objective. These dynamic weights will be adjusted in real time according to the actual operating conditions, giving priority to the more critical objectives at the moment. The final multi-objective cost function J(t) can be expressed as: Where: J1 is the goal of maximizing NOx conversion efficiency (integral form) is the goal of minimizing temperature fluctuations (temperature variance) is the goal of minimizing the radiation drift residual, w1(t), w2(t), w3(t) are dynamically adjusted weights that reflect the priority of different goals.

[0043] Specifically, the logic of optimizing the multi-objective cost function through the non-dominated sorting genetic algorithm is as follows: initialize the population: use the temperature control instruction as the genetic code to randomly generate a set of candidate solutions; fast non-dominated sorting: calculate the cost function value of each individual in the population, sort the solutions according to the cost function value, and divide the Pareto front level according to the cost function value; crowding calculation and elite retention: when performing the selection operation, calculate the crowding of the solution set and retain the elite solution to ensure the diversity and convergence of the solution set, and iteratively output the optimal temperature control instruction.

[0044] In this implementation, during the initialization phase of the genetic algorithm, the genetic representation of each individual must be set. In this problem, each individual's genetic code represents temperature control instructions (e.g., temperature setting value, adjustment method, etc.). The combination of these temperature control instructions forms a set of candidate solutions. During initialization, these control instructions are randomly generated, and the quality of each solution is unknown. This set of solutions constitutes the initial population.

[0045] In genetic algorithms, fast non-dominated sorting is a method used to evaluate the quality of each individual (i.e., each temperature control instruction). Each individual is ranked according to its cost function value, which is typically the output of a multi-objective cost function (such as NOx conversion efficiency and temperature fluctuation). The sorting method uses non-dominated sorting: for each individual in the population, its cost function value is calculated and the individuals are ranked based on these values. If a solution outperforms another solution on all objectives, it is said to "dominate" the latter. By comparing the solutions, a Pareto front is formed, which reflects the optimal solution set sought in multi-objective optimization. Non-dominated sorting divides the individuals in the population into multiple "ranks" or "fronts." Individuals in the first rank are called Pareto front solutions. These solutions are not dominated by other individuals in the population and represent the optimal solutions. Because genetic algorithms need to balance the diversity and convergence of the solution set during selection, in addition to ranking the individuals, the "crowding degree" of each solution must be calculated. Crowding degree calculation: Crowding degree measures the "sparseness" of solutions in the multi-objective space. Specifically, the proximity of each solution to its Pareto front is calculated. If a solution has fewer individuals in its neighborhood (i.e., its "crowding" level is low), it indicates greater diversity in the objective space and is likely a valid solution. Elite retention: The elitist strategy ensures that high-quality solutions from the genetic algorithm are not lost during the evolutionary process. During the selection operation, the best individuals in the current generation are retained, i.e., the "elite solutions." This ensures that excellent solutions can be passed to the next generation, enhancing the algorithm's convergence. After the above steps, the algorithm gradually generates the next generation of solutions through selection, crossover, and mutation operations. Each generation optimizes the solution set and improves solution quality through steps such as fast non-dominated sorting, crowding calculation, and elite retention. Through multiple generations of evolution, the algorithm gradually approaches the optimal solution to the multi-objective cost function. Ultimately, the algorithm outputs an optimal set of temperature control instructions, which represent the optimal solution to specific objectives (such as maximizing NOx conversion efficiency and minimizing temperature fluctuations).

[0046] In summary, this application has at least the following effects:

[0047] A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry maximizes NOx conversion efficiency and improves catalyst reaction efficiency within different temperature ranges through precise temperature control and compensation for radiation effects. By optimizing the temperature control strategy, temperature fluctuations within the reactor are reduced, preventing catalyst activity reduction or failure due to temperature instability, thereby improving system stability. Combining radiation compensation with temperature control threshold correction effectively addresses interference from external environmental factors, achieving precise temperature control and ensuring that the reaction proceeds under optimal conditions. A fuzzy logic control algorithm enables intelligent dynamic adjustment of temperature control thresholds and priority weights, automatically optimizing reaction conditions based on real-time data, improving the system's adaptability and response speed. By constructing a multi-objective cost function and optimizing it with a non-dominated sorting genetic algorithm, an optimal balance is found between maximizing NOx conversion efficiency and minimizing temperature fluctuations, improving overall system performance and stability. By accurately identifying the catalyst's activity-sensitive range and the temperature range within which catalyst deactivation is avoided, the catalyst's service life is extended and maintenance costs are reduced. The algorithm in this application can adapt to various environmental changes, particularly when radiation intensity changes, automatically adjusting the temperature control strategy to ensure that the reaction process remains efficient and stable under various conditions. This system uses big data and intelligent algorithms to process and analyze multiple data sources such as temperature, NOx concentration and radiation intensity in real time, realizing data-based dynamic optimization control.

[0048] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0049] The present invention is described with reference to flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts 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 device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0050] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0051] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0052] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0053] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A temperature control method for denitrification of low-temperature and high-NOx concentration flue gas in the nuclear industry, characterized in that: The following steps are involved: S1. Real-time acquisition of flue gas temperature time series distribution, NOx concentration gradient, and ambient radiation intensity data. A spatiotemporal alignment algorithm is used to eliminate sensor acquisition delays and generate a synchronized data matrix to characterize the coupled state of temperature, concentration, and radiation within the reactor. S2. Based on the synchronized data matrix, analyze the nonlinear impact of temperature fluctuations on NOx conversion, identify the catalyst activity-sensitive range, establish a temperature and efficiency response surface, divide the high-efficiency temperature zone into the deactivation risk temperature zone, and dynamically output the temperature control threshold and priority weight through a fuzzy logic control algorithm; S3. Correlate the response surface with the radiation intensity data to quantify the drift effect of radiation on the catalyst surface temperature. Use regression analysis to fit the functional relationship between radiation dose and temperature zone offset, modify the temperature control threshold, and generate anti-radiation compensation parameters. S4. Based on the modified temperature control threshold and anti-radiation compensation parameters, a multi-objective cost function is constructed with the goal of maximizing NOx conversion efficiency and minimizing temperature fluctuations. The multi-objective cost function is optimized using a non-dominated sorting genetic algorithm to output the optimal temperature control instruction.

2. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 1, characterized in that: The specific process of acquiring the flue gas temperature time series distribution, NOx concentration gradient change, and ambient radiation intensity data in real time, eliminating sensor acquisition delays through the spatiotemporal alignment algorithm, and generating a synchronized data matrix is ​​as follows: Based on the sensor's layout and flue gas flow rate information, the signal propagation time difference at each measurement point is calculated, and the sensor clock drift is dynamically corrected using the Kalman filter algorithm. The collected time series signals of flue gas temperature, NOx concentration and radiation intensity are aligned using a dynamic time warping algorithm; The covariance matrix of each signal is analyzed through covariance analysis to detect and eliminate abnormal data and generate a synchronous data matrix.

3. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 2, characterized in that: The specific process of analyzing the nonlinear effect of temperature fluctuations on NOx conversion and identifying the catalyst activity sensitive range based on the synchronized data matrix is ​​as follows: Based on the synchronized data matrix, the temperature and NOx concentration are mapped to a high-dimensional feature space using a Gaussian kernel function, the principal component contribution rate is calculated, the temperature and concentration correlation features are extracted, and the linear and nonlinear response components are separated. Based on the dynamic sliding window statistics, the cumulative effect of temperature fluctuations on the conversion rate is calculated. By setting different window lengths, the time response curve of temperature fluctuations to NOx conversion rate is calculated to identify the critical temperature point that causes catalyst deactivation or efficient reaction; The temperature and conversion rate distributions were clustered and analyzed using clustering algorithms to divide the catalyst activity sensitive ranges.

4. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 3, characterized in that: The specific process of establishing the temperature and efficiency response surface and dividing the high-efficiency temperature zone and the inactivation risk temperature zone is as follows: By synchronizing the temperature and NOx conversion rate data in the data matrix, a response surface of temperature and NOx conversion efficiency was established using regression analysis method; On the response surface, the effect of temperature change on NOx conversion rate is identified, and the conversion efficiency function in different temperature ranges is established; According to the conversion efficiency values ​​on the response surface, the temperature zone with higher NOx conversion rate and the deactivation risk temperature zone with lower conversion efficiency are divided; By analyzing the catalyst activity in different temperature ranges, the optimal reaction temperature range and the risk temperature range for catalyst deactivation are determined.

5. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 4, characterized in that: The specific process of dynamically outputting temperature control thresholds and priority weights through the fuzzy logic control algorithm is as follows: Based on the established response surfaces of high-efficiency temperature zone and inactivation risk temperature zone, the temperature control threshold and priority weight are dynamically output through the fuzzy logic control algorithm, including: The temperature fluctuation amplitude in the reactor and the catalyst activity state are set as input fuzzy variables, and the temperature control threshold and priority weight are set as output fuzzy variables; Set fuzzy rules to define the relationship between temperature fluctuation amplitude, catalyst activity, temperature control threshold and priority weight; Through the fuzzy control rule base, the fuzzified input variables are inferred, and the membership of each fuzzy set is comprehensively considered to obtain the corresponding temperature control threshold and priority weight.

6. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 5, characterized in that: The specific process of correlating the response surface with the radiation intensity data to quantify the effect of radiation on the catalyst surface temperature drift is as follows: Extract the temperature zone offset features in the response surface and construct a correlation matrix with the radiation intensity time series data; The time-frequency correlation between radiation and temperature drift is analyzed by cross wavelet transform; The contribution of radiation dose to the high-efficiency temperature zone boundary deviation is quantified, and a drift coefficient matrix is ​​generated.

7. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 6, characterized in that: The specific process of fitting the functional relationship between radiation dose and temperature zone offset through regression analysis, correcting the temperature control threshold, and generating anti-radiation compensation parameters is as follows: The nonlinear relationship between radiation dose and temperature zone offset was fitted by using a polynomial kernel regression model, retaining high-order nonlinear terms. In order to avoid overfitting of the regression model, elastic network regression was used to screen out radiation-related factors that have a significant impact on temperature zone shift; The regression coefficient is embedded in the threshold correction equation and the anti-radiation compensation parameter is output.

8. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 7, characterized in that: Based on the modified temperature control threshold and anti-radiation compensation parameters, with the goal of maximizing NOx conversion efficiency and minimizing temperature fluctuations, the specific process of constructing the multi-objective cost function is as follows: The area of ​​the response curve integral based on the modified temperature threshold constraint is to maximize the NOx conversion efficiency; Constrain the temperature time series variance and radiation drift residual to minimize temperature fluctuation; The dual objectives are fused through a dynamic weight distribution function to generate a multi-objective cost function.

9. A temperature control method for denitrification of low-temperature, high-NOx concentration flue gas in the nuclear industry according to claim 8, characterized in that: The logic of optimizing the multi-objective cost function through the non-dominated sorting genetic algorithm is as follows: Initialize the population: Use the temperature control instruction as the genetic code and randomly generate a set of candidate solutions; Fast non-dominated sorting: For each individual in the population, calculate its cost function value, sort the solutions according to the cost function value, and divide the Pareto front level according to the cost function value; Crowding calculation and elite retention: When performing the selection operation, the congestion of the solution set is calculated and the elite solutions are retained to ensure the diversity and convergence of the solution set, and iteratively output the optimal temperature control instructions.

Images