Intelligent control method for zinc-containing ash tail gas purification system

Abstract

The application discloses an intelligent control method for a zinc-containing ash tail gas purification system, and relates to the technical field of tail gas purification systems.The method comprises the following steps: firstly, continuously monitoring the condensation process of high-temperature zinc-containing steam in a cooler through a monitoring device, and collecting the zinc steam condensation conditions and particle generation data of the tail gas under different temperature and pressure conditions in real time.The application realizes precise control of the zinc-containing ash tail gas purification system by combining intelligent monitoring, signal analysis, machine learning identification and dynamic regulation.Through real-time monitoring and data preprocessing to extract key features, intelligent identification of particle concentration is realized based on feature vectors and machine learning, the speed of the centrifugal separator is adjusted adaptively, and pulse back-blowing dust removal is linked to ensure efficient separation and system stability.Compared with the traditional fixed speed mode, the application improves the tail gas purification efficiency, reduces the maintenance cost, prolongs the service life of the equipment, and enhances the automatic operation capability, so that the system can maintain the best working state when the load fluctuates.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tail gas purification systems, specifically to an intelligent control method for a zinc ash-containing tail gas purification system. BACKGROUND

[0002] The intelligent control of the zinc ash-containing tail gas purification system refers to the efficient and precise management of the furnace gas purification process based on automatic sensing, real-time monitoring, and intelligent adjustment, to ensure the stable operation and optimized energy efficiency of the purification system. The system integrates online monitoring devices for multiple parameters such as oxygen content, dew point, and hydrogen content, real-time collects the purification effect of furnace gas, and combines pressure sensors, frequency conversion fan controllers, and automatic ash removal systems to perform intelligent adjustment through PLC (Programmable Logic Controller) or DCS (Distributed Control System). When the system detects that the oxygen content or dew point exceeds the set threshold, it can automatically adjust the fan pressure, filter switching, or start the purge mode to avoid a decrease in purification efficiency or system blockage. In addition, the frequency conversion control of the fan can dynamically adjust the air volume according to the real-time furnace pressure fluctuations, ensuring stable furnace pressure and preventing gas leakage or excessive air supply. At the same time, the system has automatic / manual ash removal functions that can trigger the ash removal process when the filtration resistance increases, ensuring long-term stable operation. All operating states, fault alarms, and historical data can be visualized through HMI (Human Machine Interface) or remote monitoring platforms, achieving unattended operation, adaptive optimization, and intelligent operation and maintenance, and improving system reliability and production efficiency.

[0003] The existing technology has the following disadvantages:

[0004] In the existing technology, when using a centrifugal separator to remove large zinc ash particles from zinc ash-containing tail gas, a fixed rotational speed is usually used for separation, i.e., a fixed rotational speed is pre-set to separate large zinc ash particles from zinc ash-containing tail gas. However, when the concentration of zinc ash particles in the tail gas significantly increases, the fixed rotational speed cannot automatically adjust the centrifugal force in time, resulting in a large number of zinc ash particles being unable to be effectively thrown out, and thus entering the subsequent filter device. This will significantly increase the frequency of filter device blockage, causing rapid clogging and frequent replacement of filter cartridges, severely reducing the service life of the filter device, and significantly increasing equipment maintenance costs, reducing the economic efficiency and stability of the entire system operation.

[0005] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The purpose of the present application is to provide an intelligent control method for a zinc-containing ash tail gas purification system, which combines intelligent monitoring, signal analysis, machine learning identification, and dynamic regulation to achieve precise control of the zinc-containing ash tail gas purification system. By real-time monitoring and data preprocessing to extract key features, intelligent identification of particle concentration is achieved based on feature vectors and machine learning, and the speed of the centrifugal separator is adjusted adaptively, with pulse backblowing for dust removal, to ensure efficient separation and system stability. Compared with the traditional fixed speed method, the present application improves the tail gas purification efficiency, reduces the maintenance cost, prolongs the equipment life, and enhances the automation operation capability, so that the system can maintain the best working state under load fluctuations, to solve the problems in the above background technology.

[0007] To achieve the above purpose, the present application provides the following technical solution: an intelligent control method for a zinc-containing ash tail gas purification system, comprising the following steps:

[0008] Firstly, the condensation process of high-temperature zinc-containing steam in the cooler is continuously monitored by a monitoring device, and the zinc vapor condensation and particle generation data of the tail gas under different temperature and pressure conditions are collected in real time;

[0009] The obtained real-time monitoring data of zinc vapor condensation is preprocessed, and a signal analysis method is used to extract key features reflecting the rising trend of zinc ash particle concentration from the real-time data after purification;

[0010] The extracted key features are analyzed in depth to form a feature vector reflecting the current zinc ash particle concentration fluctuation state, and the dynamic fluctuation characteristics of the zinc ash particle concentration are quantified;

[0011] The generated feature vector is input into a pre-trained machine learning model to determine whether the current zinc ash particle concentration has a significant rising trend;

[0012] When the machine learning model identifies that the zinc ash particle concentration has significantly risen, the speed of the centrifugal separator is dynamically increased to increase the centrifugal force by increasing the rotor speed according to the amplitude of the particle concentration rise, and the pulse backblowing for dust removal is started to remove the zinc ash particles rapidly accumulated inside the filter unit due to the sudden rise in particle concentration by high-frequency and short-period pulse air blowing, preventing rapid clogging of the surface and interior of the filter element.

[0013] Preferably, the specific steps of collecting the zinc vapor condensation and particle generation data of the tail gas under different temperature and pressure conditions by the monitoring device are as follows:

[0014] Firstly, high-precision sensors, including but not limited to temperature sensors, pressure sensors, humidity sensors, and laser particle sensors, are arranged inside the cooler and at key pipeline positions to continuously collect the condensation data of high-temperature zinc-containing steam;

[0015] When high-temperature zinc-containing steam enters the cooler, the zinc vapor gradually condenses into solid zinc ash as the temperature gradually decreases, and the sensor records the condensation rate, particle concentration, particle diameter distribution, and gas transmittance change of the tail gas under different temperature and pressure conditions in real time;

[0016] All data are collected at high frequency by the data acquisition module, and time synchronization markers are used to ensure the time consistency of each data stream;

[0017] Finally, all monitoring data are stored in the central control system as the basis for subsequent intelligent identification of zinc ash particle concentration fluctuation trends, providing accurate data support for dynamic adjustment.

[0018] Preferably, the signal analysis method is used to extract key features reflecting the rising trend of zinc ash particle concentration from the real-time data after purification. The extracted features include the scale change of the airflow turbulence structure and the nucleation rate of the solid particles formed by steam condensation. The scale change of the airflow turbulence structure and the nucleation rate of the solid particles formed by steam condensation are analyzed under the detection window to generate turbulence vortex scale offset reference values and particle average charge change rate reference values, respectively. The dynamic fluctuation characteristics of zinc ash particle concentration are quantified by the turbulence vortex scale offset reference values and the particle average charge change rate reference values.

[0019] Preferably, the specific steps for analyzing the scale change of the airflow turbulence structure under the detection window to generate the turbulence vortex scale offset reference value are as follows:

[0020] Under the detection window, first, the turbulence structure in the airflow is decomposed using turbulence spectrum analysis and scale decomposition methods to obtain vortex features at different scales. Wavelet packet decomposition is used to decompose the turbulence velocity field, and a local scale energy function is constructed to capture the vortex change pattern at different scales in the airflow. The local scale energy distribution of the turbulence velocity vector field is calculated, and the calculation expression is as follows:

[0021]

[0022] In the formula, E s is the local turbulence energy at scale s, Ω is the spatial integration region, W n，s (x, y, z) is the nth wavelet packet decomposition coefficient at scale s in three-dimensional space coordinates (x, y, z), α is the nonlinear scaling exponent, and N is the wavelet decomposition order.

[0023] After obtaining the local turbulence energy E s at scale s, the turbulence vortex scale offset reference value is calculated to quantify the degree of turbulence scale abnormal offset. A scale energy offset function is defined to represent the contribution of scale s to the overall turbulence structure, and the calculation expression is as follows:

[0024]

[0025] where Φ(s) is the scale energy shift function, which describes the contribution of different scale s to the overall turbulent structure, S is the set of all scales, S' represents the scale range divided in the wavelet decomposition, ∑ S'∈S E S' is the total turbulent energy at all scales, E s (x, y, z) is the wavelet decomposition coefficient of scale s at three-dimensional spatial coordinates (x, y, z), is the high-order gradient of the turbulent velocity field, β is the turbulent nonlinear diffusion exponent, and γ is the exponential adjustment factor for adjusting the weight of the high-order gradient term;

[0026] Based on the scale energy shift function Φ(s), the overall turbulent vortex scale shift reference value is calculated, and the calculation expression is as follows:

[0027]

[0028] where TESSI is the turbulent vortex scale shift reference value, and ε is a small positive number to avoid mathematical singularity when the scale approaches zero.

[0029] Preferably, the nucleation rate of the steam condensing into solid particles is analyzed under the detection window to generate the specific steps of the particle average charge change rate reference value as follows:

[0030] First, the instantaneous particle collision charge transfer rate is calculated, which is used to measure the charge exchange rate of the particles caused by collision and aggregation in a high-concentration environment after the steam condenses into solid particles. The calculation expression of the instantaneous particle collision charge transfer rate is as follows:

[0031]

[0032] where Q c is the instantaneous particle collision charge transfer rate, ε0 is the vacuum permittivity, E is the local electric field strength, A p is the average surface area of the particles, is the exponential decay factor, which describes the degree of reduction of the single-particle charge exchange rate when the particle concentration N p increases, e is the natural base, and ω is the empirical adjustment parameter, K p is the particle kinetic energy, is the kinetic energy influence factor, which is used to adjust the influence degree of the kinetic energy on the charge exchange rate, F c is the Coulomb force between the particles, which describes the electrostatic interaction force between the charged particles, λ is the particle concentration influence correction factor, and θ is the adjustment parameter of the concentration exponential term;

[0033] The instantaneous particle collision charge transfer rate Qc Then, the particle average charge variation rate reference value is calculated to quantify the net charge growth trend of the particles in the entire detection window, and the calculation expression is as follows:

[0034]

[0035] In the formula, PMCVI is the particle average charge variation rate reference value, μ is the spatial area in the detection window, η is the inter-particle charge exchange efficiency factor, is the power scaling of the instantaneous particle collision charge transfer rate, δ reflects the nonlinear growth of the particle charge exchange rate, is the potential suppression factor, σ is the adjustment parameter, φ p is the particle surface potential, e is the natural base, ρ p is the particle mass density, is the particle local pressure gradient, ψ is the particle charge polarization influence factor, τ p is the particle polarization relaxation time, κ is the nonlinear scaling index of the polarization relaxation time, used to adjust the influence degree of the polarization relaxation time on the overall charge variation, is the high concentration correction factor, ξ represents the exponential influence of the particle concentration on the correction term, v is the high concentration correction factor.

[0036] Preferably, the analyzed turbulent vortex scale offset reference value and the particle average charge variation rate reference value are input into the pre-trained machine learning model, and a zinc dust particle concentration risk coefficient is generated through the machine learning model, and whether the current zinc dust particle concentration has a significant rising trend is determined through the zinc dust particle concentration risk coefficient.

[0037] Preferably, the zinc dust particle concentration risk coefficient generated when determining whether the current zinc dust particle concentration has a significant rising trend through the pre-trained machine learning model is compared with a pre-set zinc dust particle concentration risk coefficient reference threshold value, to determine whether the current zinc dust particle concentration has a significant rising trend, and the specific process is as follows:

[0038] If the zinc dust particle concentration risk coefficient is greater than the pre-set zinc dust particle concentration risk coefficient reference threshold value, it is determined that the current zinc dust particle concentration has a significant rising trend; if the zinc dust particle concentration risk coefficient is less than the pre-set zinc dust particle concentration risk coefficient reference threshold value, it is determined that the current zinc dust particle concentration does not have a rising trend.

[0039] Preferably, when the machine learning model identifies that the zinc dust particle concentration is significantly rising, the rotor speed of the centrifugal separator is dynamically increased to increase the centrifugal force according to the rising amplitude of the particle concentration, and at the same time, the pulse backblowing dust removal is started, and the specific steps are as follows:

[0040] When the machine learning model identifies a significant increase in zinc ash particle concentration in real time, the rotational speed of the centrifugal separator is immediately adjusted to enhance the centrifugal force and improve particle separation efficiency. The dynamic adjustment of the rotational speed is calculated as follows:

[0041]

[0042] where Δω is the rotational speed adjustment increment of the centrifugal separator, A is the centrifugal force gain coefficient, ZAPCR is the zinc ash particle concentration risk coefficient, ZAPCR thr is the reference threshold of the zinc ash particle concentration risk coefficient, e is the natural base, H is the non-linear adjustment index, R is the exponential adjustment parameter, κ i is the weight coefficient, indicating the contribution size of the historical time window i to the current adjustment, n is the number of historical detection windows, is the weighted historical zinc ash particle concentration risk coefficient, indicating the weighted value of the zinc ash particle concentration risk coefficient at each time in the past i time windows, and the square root index i / 2 is adjusted, δ i is the adjustment stability parameter;

[0043] After adjusting the rotational speed of the centrifugal separator, pulse backblowing is started to ensure that the zinc ash particles that have entered the filtration unit do not quickly accumulate due to sudden concentration increase, causing blockage. The frequency calculation expression of pulse backblowing is as follows:

[0044]

[0045] where f p is the adjusted pulse backblowing frequency, f p0 is the initial flow scanning frequency, E is the frequency gain adjustment coefficient, D is the adjustment smoothing coefficient, B is the exponential adjustment parameter, m is the historical detection window size, indicating the historical detection window size used to calculate the pulse jet frequency, μ j is the influence degree of the jth historical detection window on the current jet frequency calculation, is the exponential decay term, θ j is the exponential decay factor, used to control the influence degree of the past zinc ash particle concentration risk coefficient on the current pulse jet frequency, φ j is the feedback adjustment sensitivity parameter;

[0046] The backblowing gas flow pressure is calculated as follows:

[0047]

[0048] where P p is the adjusted pulse jet pressure, P p0is the initial pulse pressure, T is the pressure regulation gain coefficient, X is the non-linear regulation index, L is the pressure response sensitivity control parameter;

[0049] The centrifugal separator speed adjustment increment Δω, the adjusted pulse blowback frequency f p and the adjusted pulse blow pressure P p After calculation, a closed-loop feedback control is established to continuously optimize the centrifugal separation and pulse cleaning strategy to ensure dynamic stability under different concentration conditions. The adjustment calculation expression of real-time feedback control is as follows:

[0050]

[0051] In the formula, K adj is the adjustment coefficient, which determines the correction amount of the next cycle speed and blowback frequency, ζ is the feedback regulation gain coefficient, U is the non-linear adjustment index, which controls the change amplitude, G is the flow weight coefficient, Q is the flow, ψ f is the weight of the fth time window, v f is the smooth adjustment factor.

[0052] In the above technical solution, the technical effects and advantages provided by the present application are:

[0053] The present application realizes precise control of the zinc ash-containing tail gas purification system by combining intelligent monitoring, signal analysis, machine learning identification and dynamic regulation, and significantly improves the efficient separation capacity and system stability of zinc ash particles. First, real-time monitoring and data preprocessing technology is used to obtain and extract key features that can accurately reflect the fluctuation trend of zinc ash particle concentration, ensuring high precision and real-time performance of the data. Second, based on feature vector construction and machine learning intelligent identification, adaptive judgment of particle concentration change trend is realized, avoiding the problem that traditional fixed speed mode cannot adapt to dynamic load changes. Finally, through dynamic adjustment of centrifugal separator speed and pulse blowback linkage cleaning, when the particle concentration abnormally increases, the separation efficiency is rapidly improved and the filter unit is prevented from being blocked, thereby greatly reducing the maintenance frequency and improving the operating life and energy efficiency of the purification system. The intelligent control method of the present application effectively overcomes the problem that the fixed speed centrifugal separator in the prior art cannot be adjusted when the particle load fluctuates, making the tail gas purification system have higher automation level and operation reliability, reducing equipment maintenance cost, and improving the overall efficiency of tail gas purification. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0055] Figure 1 The method flow chart of the intelligent control method of the zinc-containing ash tail gas purification system. DETAILED DESCRIPTION

[0056] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the inventive concept to those skilled in the art.

[0057] The present application provides an intelligent control method of a zinc-containing ash tail gas purification system as shown in Figure 1 The intelligent control method of the zinc-containing ash tail gas purification system includes the following steps:

[0058] First, through the monitoring device, the condensation process of high-temperature zinc-containing steam in the cooler is continuously monitored, and the zinc steam condensation and particle generation data of the tail gas under different temperature and pressure conditions are collected in real time;

[0059] The monitoring device (such as a temperature sensor, a pressure sensor, a humidity sensor, a laser particle sensor, etc.) is used to continuously and real-time monitor the condensation process of high-temperature zinc-containing steam in the cooler, and collect the steam condensation and zinc ash particle generation data of the tail gas under different temperature and pressure conditions. In the cooler, high-temperature zinc-containing steam will gradually condense into solid zinc ash as the temperature decreases, and the condensation rate, particle size and particle concentration are affected by factors such as temperature gradient, pressure fluctuation and gas flow rate. Therefore, by real-time monitoring of these key physical parameters, the condensation dynamics of zinc steam can be accurately mastered, and the formation trend and concentration change of particles can be analyzed. The main role of this step is to establish a real-time database of tail gas particle generation, to provide high-precision and continuous original data support for subsequent intelligent identification of zinc ash particle concentration fluctuations, and to ensure that the entire tail gas purification system can accurately perceive and adapt to real-time changes in zinc ash particle concentration, thereby improving the intelligent control ability of the system.

[0060] The specific steps of collecting the zinc steam condensation and particle generation data of the tail gas under different temperature and pressure conditions by the monitoring device are as follows:

[0061] First, high-precision sensors, including temperature sensors, pressure sensors, humidity sensors, laser particle sensors, or light scattering particle concentration detectors, are placed inside the cooler and at key pipeline locations to continuously collect condensation data of high-temperature zinc-containing steam. When high-temperature zinc-containing steam enters the cooler, as the temperature gradually decreases, zinc vapor gradually condenses into solid zinc ash. The sensors will record the condensation rate, particle concentration, particle diameter distribution, and gas light transmittance changes of the tail gas under different temperature and pressure conditions in real time. All data are collected at high frequency by the data acquisition module (DAQ) and time-synchronized to ensure the time consistency of each data stream. Finally, all monitoring data are stored in the central control system as the basis for subsequent intelligent identification of zinc ash particle concentration fluctuation trends, providing accurate data support for dynamic adjustment of the system.

[0062] The acquired real-time monitoring data of zinc vapor condensation are preprocessed, and signal analysis methods such as Fourier transform, wavelet analysis, or empirical mode decomposition are used to extract key features reflecting the rising trend of zinc ash particle concentration from the purified real-time data.

[0063] The acquired real-time monitoring data of zinc vapor condensation are preprocessed, and signal analysis methods such as Fourier transform, wavelet analysis, or empirical mode decomposition are used to extract key features reflecting the rising trend of zinc ash particle concentration from the purified real-time data.

[0064] The extracted key features are analyzed in depth to form a feature vector reflecting the current zinc ash particle concentration fluctuation state, quantifying the dynamic fluctuation characteristics of zinc ash particle concentration.

[0065] The key features reflecting the rising trend of zinc dust particle concentration are extracted from the real-time data after purification by using signal analysis method, including the scale change of air flow turbulent structure and the nucleation rate of solid particles formed by steam condensation. The scale change of air flow turbulent structure and the nucleation rate of solid particles formed by steam condensation are analyzed under the detection window to generate the vortex scale deviation reference value and the average particle charge change rate reference value respectively, and the dynamic fluctuation characteristics of zinc dust particle concentration are quantified by the vortex scale deviation reference value and the average particle charge change rate reference value.

[0066] The abnormal decrease of the scale of air flow turbulent structure indicates that the concentration of zinc dust particles abnormally rises in the current condensation process of zinc dust particles, and the core reason lies in the influence of high-concentration particles on turbulent energy dissipation and the change of particle transport behavior. Under normal circumstances, the vortex scale of turbulence remains within a reasonable range, and the large-scale vortex structure can promote the uniform diffusion of zinc vapor in the air flow, enable it to condense into solid particles stably, and help the uniform distribution and settlement of particles. However, when the concentration of zinc dust particles abnormally rises, a large number of particles collide and gather with each other, and have a strong interaction with the air flow, which leads to faster dissipation of turbulent kinetic energy, rapid splitting of large-scale vortex into small-scale vortex, and finally leads to the decrease of the turbulent vortex scale of the air flow. The direct consequence of this phenomenon is the prolongation of the suspension time of particles in the air flow, the further increase of particle density in local area, and the formation of higher-concentration particle clusters, which are difficult to be efficiently thrown out by the conventional centrifugal separator, thereby increasing the burden of the subsequent filtration system. In addition, the abnormal decrease of the turbulent scale also affects the transport capacity of the air flow, making the particles in some areas more concentrated, forming a local high-concentration zinc dust area, and further aggravating the abnormal enrichment of particles. Therefore, when the turbulent vortex scale abnormally decreases, it can be used as a key feature signal of the abnormal rise of particle concentration, and the system needs to adjust the operating parameters (such as increasing the rotation speed of the centrifugal separator or optimizing the filtration mode) in time to maintain the stable operation of the system.

[0067] The specific steps of analyzing the scale change of air flow turbulent structure under the detection window to generate the vortex scale deviation reference value are as follows:

[0068] Under the detection window, first, the turbulent structure in the air flow is decomposed by using turbulent energy spectrum analysis and scale decomposition method to obtain the vortex characteristics under different scales, the wavelet packet decomposition (WPD) is used to decompose the turbulent velocity field, and the local scale energy function is constructed to capture the vortex change mode of different scales in the air flow. The local scale energy distribution of the turbulent velocity vector field is calculated, and the calculation expression is as follows:

[0069]

[0070] In the formula, E sis the local turbulent energy at scale s, Ω is the spatial integration region, used to calculate the overall turbulent energy contribution of the airflow within a specific window, W n，s (x, y, z) is the nth wavelet packet decomposition coefficient at scale s in three-dimensional spatial coordinates (x, y, z), α is a nonlinear scaling exponent, usually α > 1 is taken to emphasize the contribution of high-energy regions, making them more sensitive to abnormal turbulent changes, N is the wavelet decomposition order, representing the decomposition order used in the wavelet packet decomposition process;

[0071] The role of the above steps is to extract the turbulent energy distribution at different scales through non-uniform scale analysis, avoiding the limitations of traditional mean processing, especially suitable for capturing local energy abnormal aggregation areas. In the case of increased particle concentration, the vortex structure tends to be smaller in scale, resulting in W n，s (x, y, z) abnormally increases at small scales, ultimately making E s Significantly deviates at low scales, forming a characteristic signal of abnormal increase in particle concentration.

[0072] After obtaining the local turbulent energy E s at scale s, calculate the turbulent vortex scale shift reference value to quantify the degree of turbulent scale abnormal shift, and define the scale energy shift function to represent the contribution of scale s to the overall turbulent structure, the calculation expression is as follows:

[0073]

[0074] In the formula, Φ(s) is the scale energy shift function, used to describe the contribution of different scales s to the overall turbulent structure, measure the local energy proportion at the current scale and its high-order change characteristics, S is the collection of all scales, represents all scale ranges S' divided in the wavelet decomposition, ∑ S'∈S E S' is the total turbulent energy at all scales, represents the total turbulent energy at all scales S', as a normalization factor, so that the scale energy E s participates in the calculation in proportion, W s (x, y, z) is the wavelet decomposition coefficient at scale s in three-dimensional spatial coordinates (x, y, z), is the high-order gradient of the turbulent velocity field, representing the high-order derivative of the turbulent velocity field in the spatial x direction at scale s, used to measure the complexity of the local structure of the turbulent flow, β is the nonlinear diffusion index of the turbulent flow, which enhances the sensitivity to the change of vortex intensity, γ is the exponential adjustment factor, used to adjust the weight of the high-order gradient term to balance the contribution between different scales;

[0075] Based on the scale energy shift function Φ(s), the overall turbulent vortex scale shift reference value is calculated, and the calculation expression is as follows:

[0076]

[0077] where TESSI is the turbulent eddy scale shift reference value, and ε is a small positive number to avoid mathematical singularity when the scale approaches zero.

[0078] The greater the turbulent eddy scale shift reference value generated by analyzing the scale change of the airflow turbulent structure under the detection window, the higher the abnormal increase in the concentration of the zinc ash particles in the current condensation process. Conversely, if the reference value is smaller or stable, it indicates that the concentration of the zinc ash particles in the current condensation process is within the normal range. The core principle lies in that when the particle concentration increases, it will intensify the dissipation of turbulent energy, break the large-scale turbulent structure into more small-scale vortices, and thus cause the eddy scale to shift and tend to small-scale structure. When the turbulent eddy scale shift reference value rises, it means that the dominant eddy scale in the airflow turbulent structure significantly decreases in a short time, indicating that the particle load in the airflow increases, the mutual collision and agglomeration effect between particles are enhanced, and the particle aggregation and local concentration surge. This change will directly lead to the decrease of particle settling velocity, the increase of filtration system load, and even the problem of filter clogging. Therefore, under the monitoring window, by analyzing the scale change of the airflow turbulent structure, if the performance value of the turbulent eddy scale shift reference value significantly increases, timely control measures should be taken, such as increasing the speed of the centrifugal separator or enhancing the ash removal capacity of the filtration system, to prevent the decrease of system efficiency and the increase of maintenance cost. On the contrary, if the turbulent eddy scale shift reference value remains within the normal range, it indicates that the current zinc ash particle concentration is stable, the airflow turbulent structure has not changed abnormally, and the purification system is running well.

[0079] The sudden increase in the nucleation rate of steam condensation into solid particles usually indicates an abnormal increase in the current zinc dust particle condensation process, and the main reason is that the sharp increase in the nucleation rate means that the number of initial zinc dust particles generated per unit time is much higher than the normal level, resulting in a rapid increase in particle concentration. When the steam enters the cooler and undergoes a supercooling process, a sudden drop in local temperature or a sudden increase in steam supersaturation will cause a large amount of zinc vapor to condense rapidly in a short time, forming a high-density of tiny particles. This abnormal high nucleation rate will cause a reference value level increase in the number of particles, significantly increasing the particle load in the gas stream, and thus affecting the subsequent particle settling, transport and separation. Especially in the case of high gas flow rate or enhanced turbulence, a large number of newly formed particles cannot be effectively separated and will continue to be suspended in the gas stream, forming a high-concentration zinc dust area, ultimately increasing the burden on the downstream filter system. In addition, the particles formed at a high nucleation rate are usually small in size and large in surface area, which makes them more likely to agglomerate and form more complex particle aggregates, thereby increasing the risk of filter clogging. Therefore, when the nucleation rate suddenly increases, it means that the particle concentration has abnormally increased, and the system needs to adjust the centrifugal separator speed or optimize the cooling strategy in time to prevent excessive accumulation of zinc dust in the system and affect the purification efficiency and equipment stability.

[0080] The specific steps for analyzing the nucleation rate of steam condensation into solid particles under the detection window to generate the reference value of the average particle charge change rate are as follows:

[0081] First, calculate the instantaneous particle collision charge transfer rate, which is used to measure the charge exchange rate of particles in a high-concentration environment due to collision and aggregation after the condensation of steam into solid particles. The calculation expression of the instantaneous particle collision charge transfer rate is as follows:

[0082]

[0083] In the formula, Q c is the instantaneous particle collision charge transfer rate, ε0 is the vacuum permittivity, which represents the charge influence coefficient of particles in air medium, E is the local electric field strength, which is related to the particle charging condition, A p is the average surface area of the particles, which determines the interaction ability of the particles with the gas environment, is the exponential decay factor, which describes the degree of reduction of single particle charge exchange rate when the particle concentration N p increases, e is the natural base, ω is the empirical adjustment parameter, K p is the particle kinetic energy, which describes the kinetic energy of particles in the gas turbulence, and determines the collision frequency between particles, is the kinetic energy influence factor, which is used to adjust the influence degree of kinetic energy on the charge exchange rate, F cis the inter-particle Coulomb force, describing the electrostatic interaction between charged particles, λ is the particle concentration influence correction factor, used to correct the effect of Coulomb force in high concentration environment, θ is the concentration index term adjustment parameter, used to describe the influence of particle concentration on Coulomb force F c nonlinear relationship of the influence;

[0084] The instantaneous particle collision charge transfer rate Q c After that, the particle average charge change rate reference value is calculated to quantify the net charge growth trend of the particles within the entire detection window, and the calculation expression is as follows:

[0085]

[0086] In the formula, PMCVI is the particle average charge change rate reference value, μ is the spatial area within the detection window, that is, the effective range of the system monitoring the concentration change of zinc ash particles, η is the particle charge exchange efficiency factor, which depends on the dielectric properties of the particle material, is the power scaling of the instantaneous particle collision charge transfer rate, δ reflects the nonlinear growth of the particle charge exchange rate, is the potential suppression factor, σ is the adjustment parameter, φ p is the particle surface potential, reflecting the adsorption / rejection effect of the self-charged particles on the newly generated charges, e is the natural base, ρ p is the particle mass density, is the particle local pressure gradient, describing the uneven distribution of particles in the gas flow, ψ is the particle charge polarization influence factor, τ p is the particle polarization relaxation time, indicating the time required for the charge to reach a stable state after being affected by an external electric field, κ is the nonlinear scaling index of the polarization relaxation time, used to adjust the influence degree of the polarization relaxation time on the overall charge change, is the high concentration correction factor, ξ represents the exponential influence of particle concentration on the correction term, v is the high concentration correction factor.

[0087] The greater the particle average charge change rate reference value generated after analyzing the nucleation rate of steam condensation into solid particles under the detection window, the higher the abnormal increase in the concentration of zinc ash particles in the current condensation process, and vice versa. This is because in the process of steam condensation into solid particles, particles will accumulate and transfer charges due to factors such as collision, adhesion, and surface interaction. When the nucleation rate abnormally increases, the number of particles generated in a short period of time increases dramatically, resulting in a significant increase in the opportunity for particle collision and agglomeration, thereby accelerating the charge exchange on the surface of the particles, causing the particle average charge change rate reference value to rise sharply. In a high particle concentration environment, the electrostatic repulsion between particles is enhanced, making it more difficult for them to settle and more likely to be suspended in the airflow, further exacerbating the burden on the downstream filtration system.

[0088] The generated feature vector is input into a pre-trained machine learning model to determine whether the current zinc dust particle concentration shows a significant upward trend;

[0089] The analyzed turbulent vortex scale offset reference value and particle average charge change rate reference value are input into a pre-trained machine learning model to generate a zinc dust particle concentration risk coefficient, and the zinc dust particle concentration risk coefficient is used to determine whether the current zinc dust particle concentration shows a significant upward trend.

[0090] The pre-trained machine learning model refers to a model that is trained offline before actual operation through a large amount of historical data, experimental data, and simulation data. This model can learn and identify the change pattern of zinc dust particle concentration and intelligently predict and classify real-time input data during actual production. In this scheme, the turbulent vortex scale offset reference value and the particle average charge change rate reference value are two important feature parameters that can reflect the changes in airflow turbulence structure and the electrostatic interaction between particles. These factors are crucial for the condensation, aggregation, and separation of zinc dust particles. However, analyzing the changes of a single parameter alone is often insufficient to accurately determine whether the particle concentration is abnormally high. Therefore, it is necessary to use a machine learning model to correlate multiple feature parameters and generate a zinc dust particle concentration risk coefficient through pattern recognition technology to achieve more accurate concentration change prediction. The main function of the pre-trained machine learning model is to learn the rules of particle concentration increase and construct a multi-dimensional feature space of concentration change. In actual operation, it can identify the trend of concentration anomalies in real time and efficiently, and provide decision support so that the system can take dynamic adjustment measures, such as adjusting the speed of the centrifugal separator or triggering the pulse blowback mode.

[0091] To build an efficient and accurate zinc ash particle concentration prediction model, a large amount of historical monitoring data needs to be collected first, including data under normal operating conditions and data when the concentration abnormally rises, to ensure that the model can cover all possible operating conditions. These data usually contain multiple variables, such as turbulent vortex scale offset reference value, particle average charge change rate reference value, gas light transmittance decay rate, particle aggregation rate, etc., and need to be preprocessed such as data cleaning, noise reduction, standardization, etc. to improve data quality. Then, select appropriate machine learning algorithms for training, such as support vector machine (SVM), random forest, long short-term memory neural network (LSTM), or deep neural network (DNN), and continuously optimize model parameters to improve prediction accuracy. During training, the model learns the trend of zinc ash particle concentration, extracts hidden features, and establishes a mapping relationship between input parameters and concentration risk coefficients. After training, the model is deployed to the real-time monitoring system and uses real-time input turbulent vortex scale offset reference value and particle average charge change rate reference value data for dynamic prediction, calculating the zinc ash particle concentration risk coefficient. If the risk coefficient exceeds the set safety threshold, it indicates that the particle concentration is rising significantly, and the system needs to take appropriate control measures to prevent filter clogging or tail gas purification system failure. Through this pre-trained machine learning model, the intelligent level of zinc ash particle concentration monitoring can be greatly improved, achieving accurate prediction and dynamic optimization control to ensure long-term stable and efficient operation of the system.

[0092] The machine learning model is not limited here, and any machine learning model that can analyze the turbulent vortex scale offset reference value TESSI and the particle average charge change rate reference value PMCVI to generate the zinc ash particle concentration risk coefficient ZAPCR can be used. To achieve the technical solution of the present application, a specific implementation scheme is provided.

[0093] The zinc ash particle concentration risk coefficient ZAPCR generation formula is as follows: ZAPCR = k1·TESSI + k2·PMCVI, where k1 and k2 are the preset proportion coefficients of the turbulent vortex scale offset reference value TESSI and the particle average charge change rate reference value PMCVI, and both k1 and k2 are greater than 0.

[0094] The preset proportion coefficients refer to the weighting factors for measuring the influence degree of the turbulent vortex scale shift reference value TESSI and the particle average charge variation rate reference value PMCVI on the zinc ash particle concentration risk coefficient ZAPCR, that is, k1 and k2 in the formula. The main role of these coefficients is to determine the weight of the turbulent vortex scale shift reference value TESSI and the particle average charge variation rate reference value PMCVI in the calculation of ZAPCR according to historical data, experimental results or machine learning trained models, to ensure that the risk assessment can more accurately reflect the real changes of particle concentration. For example, if the influence of turbulent structure change on particle concentration is greater, the value of k1 should be higher, and vice versa, if the influence of charge change on particle agglomeration behavior is more significant, k2 should be dominant. These preset proportion coefficients are usually determined by regression analysis, optimization algorithm or neural network training, and may be dynamically adjusted according to different working conditions to optimize the prediction accuracy and response ability of the system.

[0095] From the zinc ash particle concentration risk coefficient, the greater the turbulent vortex scale shift reference value generated by analyzing the scale change of the turbulent structure of the gas flow under the detection window, the greater the particle average charge variation rate reference value generated by analyzing the nucleation rate of the steam condensation to form solid particles under the detection window, the greater the zinc ash particle concentration risk coefficient generated by the pre-trained machine learning model when judging whether the current zinc ash particle concentration has a significant rising trend, indicating that the current zinc ash particle concentration abnormally rises during the condensation process, and vice versa, indicating that the current zinc ash particle concentration is stable.

[0096] The zinc ash particle concentration risk coefficient generated by the pre-trained machine learning model when judging whether the current zinc ash particle concentration has a significant rising trend is compared with the pre-set zinc ash particle concentration risk coefficient reference threshold to determine whether the current zinc ash particle concentration has a significant rising trend, and the specific process is as follows:

[0097] If the zinc ash particle concentration risk coefficient is greater than the pre-set zinc ash particle concentration risk coefficient reference threshold, it is determined that the current zinc ash particle concentration has a significant rising trend; if the zinc ash particle concentration risk coefficient is less than the pre-set zinc ash particle concentration risk coefficient reference threshold, it is determined that the current zinc ash particle concentration does not have a rising trend.

[0098] When the machine learning model identifies that the zinc ash particle concentration is significantly rising in real time, according to the amplitude of the particle concentration rising, the speed of the centrifugal separator is dynamically adjusted to increase the rotor speed to increase the centrifugal force, at the same time, the pulse back blowing ash is started, through high frequency, short cycle pulse air blowing, the zinc ash particles rapidly accumulated inside the filter unit due to the sudden rise of particle concentration are removed, to prevent the rapid clogging of the surface and inside of the filter element;

[0099] When the concentration of zinc ash particles abnormally increases, the system's rapid response is achieved by dynamically adjusting the centrifugal separator speed and pulse backblowing to ensure effective particle separation, prevent the filtration unit from overloading, and maintain the stable operation of the entire purification system. When the machine learning model detects that the particle concentration exceeds the set threshold, it means that the large accumulation of zinc ash particles in the exhaust gas has exceeded the normal operating range. If the separation and dust removal strategy is not adjusted in time, it may cause rapid clogging of the subsequent filtration system, increase the system resistance, and ultimately affect the exhaust gas purification efficiency or even cause system failure. Therefore, this step first dynamically adjusts the centrifugal separator speed according to the magnitude of the particle concentration increase, adjusts the rotor speed using a frequency drive to make the particles in the gas flow subject to a stronger centrifugal force, ensuring that more large particles are effectively thrown out and reducing the dust load entering the filtration unit. At the same time, the pulse backblowing mode is triggered to quickly clean the zinc ash particles inside the filtration unit by high-frequency, short-period pulse airflow blowing, causing the particles adhering to the filter surface to fall off, preventing particles from accumulating inside the filtration unit, and avoiding filter clogging. The core purpose of this step is to build an intelligent adaptive adjustment mechanism to ensure that the system can automatically optimize the operating state when the particle concentration suddenly changes, improve purification efficiency, and reduce equipment maintenance costs and downtime risks. Through the linkage control of centrifugal separator speed and pulse backblowing, the exhaust gas treatment system is always in an efficient and stable working state, effectively preventing equipment damage, filtration failure, or exhaust emission exceeding the standard caused by sudden particle concentration increase.

[0100] When the machine learning model identifies a significant increase in zinc ash particle concentration in real time, the centrifugal separator speed is dynamically adjusted according to the magnitude of the particle concentration increase to increase the centrifugal force, and the pulse backblowing is started to remove the zinc ash particles rapidly accumulated inside the filtration unit by high-frequency, short-period pulse airflow blowing. The specific steps are as follows:

[0101] When the machine learning model identifies a significant increase in zinc ash particle concentration in real time, the centrifugal separator speed is immediately adjusted to increase the centrifugal force and improve particle separation efficiency. The dynamic adjustment of the rotational speed is calculated as follows:

[0102]

[0103] In the formula, Δω is the centrifugal separator speed adjustment increment, representing the rotational speed that the centrifugal separator needs to increase after detecting an abnormal increase in zinc ash particle concentration. A is the centrifugal force gain coefficient, used to adjust the overall magnitude of Δω, determining the maximum response capability of the speed adjustment. ZAPCR is the zinc ash particle concentration risk coefficient, ZAPCR thris the zinc ash particle concentration risk factor reference threshold, e is the natural base, H is the non-linear adjustment index, the sensitivity of the control system to the change of zinc ash concentration, makes the adjustment process more smooth, R is the exponential adjustment parameter, used to control the non-linear relationship in the calculation of Δω, affects the adjustment curvature of the rotating speed, κ i is the weight coefficient, indicating the contribution size of the historical time window i to the current adjustment, n is the number of historical detection windows, used to calculate the number of data windows that affect the history, indicating how long the system considers the past trend of zinc ash particle concentration change, is the weighted historical zinc ash particle concentration risk factor, indicating the weighted value of the zinc ash particle concentration risk factor at each time in the past i time windows, and is adjusted by the square root index i / 2, δ i is the adjustment stability parameter, to prevent excessive surge of the adjustment amount;

[0104] Effects:

[0105] Effectively improve the rotating speed of the centrifugal separator in a short time, enhance the particle separation capacity; reduce the probability of large particles entering the subsequent filtering device, and reduce the risk of blockage.

[0106] After adjusting the rotating speed of the centrifugal separator, start the pulse back-blowing dust removal at the same time, to ensure that the zinc ash particles that have entered the filtering unit will not be quickly accumulated due to sudden increase in concentration, causing blockage. The frequency calculation expression of pulse back-blowing is as follows:

[0107]

[0108] In the formula, f p is the adjusted pulse back-blowing frequency, f p0 is the initial flow scanning frequency, E is the frequency gain adjustment coefficient, D is the adjustment smoothing coefficient, to prevent excessive fluctuation in a short time, B is the exponential adjustment parameter, to control the non-linearity of frequency growth, m is the historical detection window size, indicating the historical detection window size used to calculate the pulse jet frequency, i.e. when calculating the current jet frequency, consider the zinc ash particle concentration data of the past m time steps, μ j is the influence degree of the jth historical detection window on the calculation of the current jet frequency, is the exponential decay term, used to exponentially decay the historical zinc ash particle concentration risk factor , so that the influence of the more distant concentration data on the current decision gradually decreases, θ j is the exponential decay factor, used to control the influence degree of the past zinc ash particle concentration risk factor on the current pulse jet frequency, φ j is the feedback adjustment sensitivity parameter, to control the response speed of the jet frequency to the historical concentration change;

[0109] The pulse backflow pressure is calculated, and the calculation expression is as follows:

[0110]

[0111] In the formula, P p is the adjusted pulse backflow pressure, P p0 is the initial backflow pressure, T is the pressure adjustment gain coefficient, X is the non-linear adjustment index, and L is the pressure response sensitivity control parameter;

[0112] Effects:

[0113] The pulse backflow frequency is adaptively adjusted according to the concentration increase amplitude, and the dust removal efficiency is improved.

[0114] The backflow pressure is appropriately increased to enhance the cleaning effect and avoid rapid deposition of particles in the filter unit.

[0115] The system can maintain long-term stable operation under high concentration load.

[0116] The centrifugal separator speed adjustment increment Δω, the adjusted pulse backflow frequency f p and the adjusted pulse backflow pressure P p are calculated, and a closed-loop feedback control is established to continuously optimize the centrifugal separation and pulse dust removal strategy to ensure dynamic stability under different concentration conditions. The adjustment calculation expression of real-time feedback control is as follows:

[0117]

[0118] In the formula, K adj is the adjustment coefficient, which determines the correction amount of the next cycle speed and backflow frequency, ζ is the feedback adjustment gain coefficient, U is the non-linear adjustment index, which controls the change amplitude, G is the flow weight coefficient, Q is the flow, ψ f is the weight of the fth time window, which determines the contribution degree of the historical data in the current calculation, v f is the smoothing adjustment factor, which controls the smoothing influence of the historical data on the current concentration change to avoid excessive adjustment.

[0119] Effects:

[0120] A closed-loop regulation system is formed to ensure that the centrifugal separator and the backflow system can be dynamically and adaptively optimized when the particle concentration changes.

[0121] The entire purification system has both rapid response capability and avoids instability caused by excessive adjustment.

[0122] Further reduce equipment wear and tear, improve purification efficiency, and reduce maintenance costs.

[0123] The present application realizes accurate control of the zinc ash-containing tail gas purification system by the method of intelligent monitoring, signal analysis, machine learning identification and dynamic regulation, and significantly improves the efficient separation ability of zinc ash particles and the system stability. First, real-time monitoring and data preprocessing technology is used to obtain and extract key features that can accurately reflect the fluctuation trend of zinc ash particle concentration, ensuring high accuracy and real-time of the data. Second, based on feature vector construction and machine learning intelligent identification, adaptive judgment of particle concentration change trend is realized, avoiding the problem that the traditional fixed speed mode cannot adapt to dynamic load changes. Finally, through dynamic adjustment of the centrifugal separator speed and pulse backflush linkage dust removal, when the particle concentration abnormally increases, the separation efficiency is rapidly improved and the filter unit is prevented from being blocked, thereby greatly reducing the maintenance frequency and improving the operating life and energy efficiency of the purification system. The intelligent control method of the present application effectively overcomes the problem that the fixed speed centrifugal separator in the prior art cannot adjust when the particle load fluctuates, making the tail gas purification system have higher automation level and operation reliability, reducing the equipment maintenance cost, and improving the overall efficiency of the tail gas purification.

[0124] The above formulas are dimensionless values calculated, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the most recent real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0125] The above only describes some exemplary embodiments of the present application by way of illustration, and it is needless to say that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above drawings and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

[0126] It should be noted that in this paper, if there are relationship terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment. Without more limitation, the element defined by the sentence "including a" does not exclude the presence of other identical elements in the process, method, article or equipment including the element.

[0127] It should be understood that the size of the sequence number of the above processes does not mean the order of execution in various embodiments of the present application, and the execution order of the processes should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0128] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0129] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0130] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment according to actual needs.

[0131] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0132] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0133] The above only describes certain exemplary embodiments of the present application by way of illustration, and it is self-evident that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above figures and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

Claims

1. An intelligent control method for a zinc-containing ash exhaust gas purification system, characterized in that, Includes the following steps: First, the condensation process of high-temperature zinc-containing vapor in the cooler is continuously monitored through the monitoring device, and the zinc vapor condensation and particle generation data of the exhaust gas under different temperature and pressure conditions are collected in real time. The acquired real-time monitoring data of zinc vapor condensation was preprocessed, and signal analysis methods were used to extract key features reflecting the increasing trend of zinc ash particle concentration from the purified real-time data. The key features extracted are analyzed in depth to form a feature vector that reflects the current fluctuation state of zinc ash particle concentration, thereby quantifying the dynamic fluctuation characteristics of zinc ash particle concentration. The generated feature vector is input into a pre-trained machine learning model to determine whether the current concentration of zinc ash particles shows a significant upward trend. When the machine learning model identifies a significant increase in zinc ash particle concentration in real time, it dynamically increases the centrifugal separator speed to increase the rotor speed and centrifugal force based on the magnitude of the increase in particle concentration. At the same time, it starts pulse backflushing to remove zinc ash particles that have accumulated rapidly inside the filter unit due to the sudden increase in particle concentration through high-frequency, short-cycle pulsed airflow, thus preventing rapid clogging of the filter element surface and interior. Signal analysis methods were used to extract key features reflecting the increasing trend of zinc ash particle concentration from real-time data after purification. The extracted features included the scale changes of the airflow turbulence structure and the nucleation rate of solid particles formed by steam condensation. The scale changes of the airflow turbulence structure and the nucleation rate of solid particles formed by steam condensation were analyzed under the detection window to generate reference values ​​for turbulent eddy scale offset and particle average charge change rate. The dynamic fluctuation characteristics of zinc ash particle concentration were quantified by the reference values ​​for turbulent eddy scale offset and particle average charge change rate. The specific steps for analyzing the scale changes of airflow turbulence structures within a detection window to generate reference values ​​for turbulent eddy scale migration are as follows: Within the detection window, the turbulent structure in the airflow is first decomposed using turbulent energy spectrum analysis and scale decomposition methods to obtain vortex characteristics at different scales. Wavelet packet decomposition is then used to decompose the turbulent velocity field, and a local scale energy function is constructed to capture vortex variation patterns at different scales in the airflow. The local scale energy distribution of the turbulent velocity vector field is calculated, and the calculation expression is as follows: In the formula, E s The local turbulent energy is at scale s, Ω is the spatial integral region, and W is the local turbulent energy. n，s (x, y, z) are the nth wavelet packet decomposition coefficients of scale s in three-dimensional spatial coordinates (x, y, z), α is the nonlinear scaling exponent, and N is the wavelet decomposition order; Obtaining the local turbulent energy E at scale s s Next, the reference value for turbulent eddy scale migration is calculated to quantify the degree of turbulent scale anomaly migration. The contribution of scale s to the overall turbulent structure is characterized by defining a scale energy migration function, and the calculation is expressed as follows: In the formula, Φ(s) is the scale energy shift function, used to describe the contribution of different scales s to the overall turbulent structure, S is the set of all scales, representing all scale ranges S' divided in wavelet decomposition, and Σ S'∈S E S' It is the total turbulent energy across all scales, representing the total turbulent energy at all scales S', W s (x, y, z) are the wavelet decomposition coefficients of scale s in three-dimensional spatial coordinates (x, y, z). γ is the higher-order gradient of the turbulent velocity field, β is the turbulent nonlinear diffusion exponent, and γ is the exponent adjustment factor used to adjust the weight of the higher-order gradient term. Based on the scale energy migration function Φ(s), the overall turbulent eddy scale migration reference value is calculated, and the calculation expression is as follows: In the formula, TESSI is the reference value for turbulent eddy scale offset, and ε is a small positive number to avoid mathematical singularities when the scale approaches zero.

2. The intelligent control method for the zinc-containing ash tail gas purification system according to claim 1, characterized in that, The specific steps for collecting real-time data on zinc vapor condensation and particle formation in exhaust gas under different temperatures and pressures using a monitoring device are as follows: First, high-precision sensors, including but not limited to temperature sensors, pressure sensors, humidity sensors, and laser particle sensors, are installed inside the cooler and at key pipe locations to continuously collect condensation data of high-temperature zinc-containing vapor. When high-temperature zinc-containing vapor enters the cooler, as the temperature gradually decreases, the zinc vapor gradually condenses into solid zinc ash. The sensor will record in real time the changes in condensation rate, particle concentration, particle diameter distribution and gas transmittance of the exhaust gas under different temperature and pressure conditions. All data is collected at high frequency through the data acquisition module, and time synchronization markers are used to ensure the time consistency of each data stream; Ultimately, all monitoring data is stored in the central control system, serving as the basis for subsequent intelligent identification of zinc ash particle concentration fluctuation trends and providing precise data support for dynamic adjustment.

3. The intelligent control method for the zinc-containing ash tail gas purification system according to claim 1, characterized in that, The specific steps for analyzing the nucleation rate of solid particles formed by steam condensation within a detection window to generate a reference value for the average charge change rate of the particles are as follows: First, the instantaneous particle collision charge transfer rate is calculated. This rate measures the charge exchange rate caused by collisions and aggregation of particles in a high-concentration environment after steam condenses into solid particles. The expression for calculating the instantaneous particle collision charge transfer rate is as follows: In the formula, Q c ε0 is the instantaneous charge transfer rate of particle collisions, ε0 is the vacuum permittivity, E is the local electrostatic field strength, and A is the instantaneous charge transfer rate of particle collisions. p It is the average surface area of ​​the particles. It is an exponential decay factor, describing the decay of particle concentration N as the particle concentration decreases. p When the value is increased, the degree of decrease in the charge exchange rate of a single particle is given by e, where e is the natural base, ω is an empirical adjustment parameter, and K is the value of K. p It is particle kinetic energy. It is the kinetic energy influence factor, used to adjust the degree of influence of kinetic energy on the charge exchange rate, F c λ is the Coulomb force between particles, describing the electrostatic force between charged particles; λ is the particle concentration influence correction factor; and θ is the adjustment parameter of the concentration exponent term. Obtain the instantaneous particle collision charge transfer rate Q c Next, a reference value for the average charge change rate of the particles is calculated to quantify the net charge growth trend of the particles throughout the detection window. The calculation expression is as follows: In the formula, PMCVI is the reference value for the average charge change rate of particles, μ is the spatial region within the detection window, and η is the charge exchange efficiency factor between particles. It is a power-scale of the instantaneous particle collision charge transfer rate, where δ reflects the nonlinear growth of the particle charge exchange rate. It is the potential suppression factor, σ is the adjustment parameter, and φ is the voltage suppression factor. p Let ρ be the surface potential of the particle, e be the natural base, and ρ be the surface potential of the particle. p It is the particle mass density. τ is the local pressure gradient of the particle, ψ is the particle charge polarization influencing factor, and τ is the local pressure gradient of the particle. p κ is the particle polarization relaxation time, and κ is the nonlinear scaling exponent of the polarization relaxation time, used to adjust the degree of influence of the polarization relaxation time on the overall charge change. ξ is the high concentration correction factor, which represents the exponential effect of particle concentration on the correction term, and v is the high concentration correction factor.

4. The intelligent control method for the zinc-containing ash tail gas purification system according to claim 1, characterized in that, The analyzed reference values ​​of turbulent eddy scale offset and particle average charge change rate are input into a pre-trained machine learning model. The machine learning model generates a zinc ash particle concentration risk coefficient, which is used to determine whether the current zinc ash particle concentration shows a significant upward trend.

5. The intelligent control method for the zinc-containing ash tail gas purification system according to claim 4, characterized in that, The risk coefficient of zinc ash particle concentration generated by a pre-trained machine learning model when judging whether the current zinc ash particle concentration shows a significant upward trend is compared with a pre-set reference threshold for the risk coefficient of zinc ash particle concentration to determine whether the current zinc ash particle concentration shows a significant upward trend. The specific process is as follows: If the risk coefficient of zinc ash particle concentration is greater than the preset reference threshold for zinc ash particle concentration risk coefficient, it is determined that the current zinc ash particle concentration shows a significant upward trend; if the risk coefficient of zinc ash particle concentration is less than the preset reference threshold for zinc ash particle concentration risk coefficient, it is determined that the current zinc ash particle concentration does not show an upward trend.

6. The intelligent control method for the zinc-containing ash tail gas purification system according to claim 5, characterized in that, When the machine learning model identifies a significant increase in zinc ash particle concentration in real time, it dynamically increases the centrifugal separator speed to increase the rotor speed and centrifugal force based on the magnitude of the increase. At the same time, it initiates pulse backflushing cleaning, using high-frequency, short-cycle pulsed airflow to remove the zinc ash particles that rapidly accumulate inside the filter unit due to the sudden increase in particle concentration. The specific steps are as follows: When the machine learning model detects a significant increase in zinc ash particle concentration in real time, it immediately adjusts the centrifuge's rotation speed to enhance centrifugal force and improve particle separation efficiency. The dynamic adjustment calculation expression is as follows: In the formula, Δω is the centrifuge speed adjustment increment, A is the centrifugal force gain coefficient, and ZAPCR is the zinc ash particle concentration risk coefficient. thr This is the reference threshold for the risk coefficient of zinc ash particle concentration, where e is the natural base, H is the nonlinear adjustment exponent, R is the exponential adjustment parameter, and κ is the reference threshold. i is the weighting coefficient, representing the contribution of historical time window i to the current adjustment, and n is the number of historical detection windows. This is the weighted historical zinc ash particle concentration risk coefficient, representing the weighted value of the zinc ash particle concentration risk coefficient at each moment within the past i time windows, adjusted using the square root exponent i / 2, δ. i It involves adjusting stability parameters; After adjusting the centrifuge speed, simultaneously start pulse backflushing to ensure that zinc ash particles that have entered the filter unit do not accumulate rapidly due to a sudden increase in concentration, causing blockage. The formula for calculating the pulse backflushing frequency is as follows: In the formula, f p It is the adjusted pulse backflush frequency, f p0 The initial flow scanning frequency is E, the frequency gain adjustment coefficient is D, the smoothing coefficient is B, the exponential adjustment parameter is m, and the historical detection window size is μ, representing the size of the historical detection window used to calculate the pulse jet frequency. j This represents the degree of influence of the j-th historical detection window on the calculation of the current injection frequency. It is an exponentially decaying term, θ j It is an exponential decay factor used to control the risk coefficient of past zinc ash particle concentration. The degree of influence on the current pulse jet frequency, φ j It is a feedback adjustment sensitivity parameter; The backflush air pressure is calculated using the following expression: In the formula, P p It is the adjusted pulse jet pressure, P p0 is the initial injection pressure, T is the pressure regulation gain coefficient, X is the nonlinear regulation index, and L is the pressure response sensitivity control parameter. Complete the centrifuge speed adjustment increment Δω and the adjusted pulse backflush frequency f p And the adjusted pulse jet pressure P p After calculation, a closed-loop feedback control is established to continuously optimize the centrifugal separation and pulse cleaning strategies to ensure dynamic stability under different concentration conditions. The adjustment calculation expression for real-time feedback control is as follows: In the formula, K adj ζ is the adjustment coefficient, which determines the correction amount for the rotational speed and backflush frequency in the next cycle; ζ is the feedback adjustment gain coefficient; U is the nonlinear adjustment index, which controls the magnitude of change; G is the flow weight coefficient; Q is the flow rate; and ψ is the feedback adjustment gain coefficient. f It is the weight of the f-th time window, υ f It is a smoothing adjustment factor.

Images