Intelligent control system for high-temperature flue gas treatment based on wet-type electric dust collector

Abstract

The present application belongs to the technical field of industrial flue gas purification, in particular to a high-temperature flue gas treatment intelligent control system based on a wet-type electric dust collector, comprising an information acquisition module, a data processing module, a spraying adjustment module and an execution feedback module, through machine learning mapping of pollution parameter trend combination and blockage level, accurate blockage level determination is realized, through plate blockage distribution law condition tracing of main blockage causes, the limitation that real-time blockage of the equipment cannot be accurately understood under the existing mode is broken, the accurate determination and tracing provide accurate basis for subsequent adjustment of the spraying strategy, and combined with dynamic spraying adjustment strategy, spraying parameters are quickly adjusted, the hysteresis of traditional manual trial and error is avoided, the response rate of the equipment under various conditions is improved, on-demand cleaning is realized, the problem that water waste and insufficient cleaning effect coexist is effectively solved, and thus the overall purification efficiency and operation economy of the equipment are improved.

Description

Technical Field

[0001] This invention belongs to the field of industrial flue gas purification technology, specifically a high-temperature flue gas treatment intelligent control system based on a wet electrostatic precipitator. Background Technology

[0002] A wet electrostatic precipitator is a highly efficient flue gas purification device, mainly used to capture fine particulate matter, aerosols, acid mist, heavy metals, and other pollutants in industrial flue gas. It is widely used in waste gas treatment in industries such as power, steel, and chemicals. Its core principle is to use a high-voltage electric field to charge the pollutants, then use the electric field force to adsorb them onto the collecting electrode (anode plate, hereinafter referred to as the electrode plate). Finally, water is used to flush the electrode plate to remove the pollutants, thus achieving flue gas purification.

[0003] In the existing technology, wet electrostatic precipitators have the following technical defects when washing the electrode plates through a spray system:

[0004] 1. The spraying operation of existing wet electrostatic precipitators adopts a fixed cycle and fixed amplitude mode, and fails to establish a dynamic correlation mechanism with the real-time degree and cause of equipment blockage. As a result, the spraying strategy cannot be flexibly adjusted according to the actual pollution status of the plates, ultimately resulting in an inefficient operation with both water waste and insufficient cleaning effect, which affects the overall purification efficiency and operating economy of the equipment.

[0005] 2. Existing wet electrostatic precipitators lack an iterative optimization mechanism based on operational feedback and flexible amplitude adjustment. Fixed amplitude adjustment is prone to exceeding the actual required cleaning intensity range, ultimately resulting in excessive investment and ineffective consumption of spraying resources. Summary of the Invention

[0006] To overcome the shortcomings in the background art, the present invention provides an intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator, which can effectively solve the problems involved in the background art.

[0007] The objective of this invention can be achieved through the following technical solution: a high-temperature flue gas treatment intelligent control system based on a wet electrostatic precipitator, characterized in that it includes: an information acquisition module, a data processing module, a spray adjustment module, and an execution feedback module.

[0008] The information acquisition module is connected to the data processing module, the data processing module is connected to the spray adjustment module, and the spray adjustment module is connected to the execution feedback module.

[0009] The information acquisition module monitors the pollution parameters of the flue gas purified by the outlet flue of the wet electrostatic precipitator and the wastewater from the liquid collection device. The pollution parameters include the concentration of heavy metal pollutants and the concentration of chemical oxygen demand.

[0010] The data processing module performs trend feature analysis on the pollution parameters to obtain trend phenomenon combinations, inputs them into a pre-trained machine learning model to determine the plate blockage level, and traces the main causes of blockage based on the plate blockage distribution pattern.

[0011] The spray adjustment module dynamically adjusts the spray parameters according to the plate blockage level and the main cause of blockage to perform spray adjustment. The spray parameters include spray frequency and spray pressure.

[0012] The execution feedback module analyzes the spray performance within a preset period after spray adjustment. If the effect meets the processing standard, the current spray parameters are automatically saved as the optimization benchmark value. If not, the spray parameters are adjusted according to the gradient until they meet the processing standard.

[0013] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0014] (1) This invention achieves accurate clogging level determination by combining pollution parameter trends with machine learning mapping of clogging level, and traces the main clogging causes by the distribution pattern of plate clogging. This breaks the limitation of not being able to accurately understand the real-time clogging situation of equipment under the traditional fixed mode. This accurate determination and tracing provides an accurate basis for the subsequent adjustment of spraying strategy.

[0015] (2) Based on the level and cause of blockage, the present invention combines dynamic spray adjustment strategy to quickly adjust spray parameters, avoid the lag of existing manual trial and error, improve the response rate of equipment under various conditions, realize on-demand cleaning, effectively solve the problem of water waste and insufficient cleaning effect, thereby improving the overall purification efficiency and operating economy of the equipment.

[0016] (3) The present invention reflects whether the spraying adjustment is appropriate by the spraying effect, and then iterates to the optimal spraying parameter combination by gradient adjustment. This iterative optimization mechanism based on operation feedback and flexible amplitude adjustment avoids the problem of excessive investment and ineffective consumption caused by fixed amplitude adjustment, and reduces operating costs. By continuously providing feedback and adjustment based on the spraying effect, closed-loop automated operation is achieved. Attached Figure Description

[0017] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the module connection of the present invention.

[0019] Figure 2 This is a flowchart illustrating the logical process of pollution parameter trend characteristic analysis in this invention.

[0020] Figure 3 This is a diagram illustrating the construction process of the machine learning model of this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figure 1 As shown, the present invention provides an intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator, including: an information acquisition module, a data processing module, a spray adjustment module, and an execution feedback module.

[0023] The information acquisition module is connected to the data processing module, the data processing module is connected to the spray adjustment module, and the spray adjustment module is connected to the execution feedback module.

[0024] The information acquisition module is used to monitor the pollution parameters of the flue gas purified by the outlet flue of the wet electrostatic precipitator and the wastewater of the liquid collection device. The pollution parameters include the concentration of heavy metal pollutants and the concentration of chemical oxygen demand.

[0025] It should be added that online metal analyzers are installed in both the outlet flue of the wet electrostatic precipitator and the liquid collection device to obtain the concentration of heavy metals in the flue gas and wastewater, and COD sensors are installed in the liquid collection device to obtain the concentration of chemical oxygen demand.

[0026] Laser-induced breakdown spectroscopy is installed in the outlet flue of the wet electrostatic precipitator to collect the content of particulate matter such as lead and cadmium compounds in real time to reflect the concentration of heavy metal pollutants. Inductively coupled plasma mass spectrometry is installed in the liquid collection device to analyze the concentration of heavy metals such as mercury, lead and cadmium in the wastewater. An electrochemical COD analyzer is also installed in the liquid collection device to measure the total amount of organic matter and inorganic reducing substances to reflect the concentration of chemical oxygen demand.

[0027] The data processing module performs trend feature analysis on the pollution parameters to obtain trend phenomenon combinations, inputs them into a pre-trained machine learning model to determine the plate blockage level, and traces the main causes of blockage based on the plate blockage distribution pattern.

[0028] Reference Figure 2 As shown, in a preferred embodiment of the present invention, the specific processing procedure for trend characteristic analysis of the pollution parameters includes the following steps:

[0029] Within a preset time period, the concentrations of heavy metals in the purified flue gas from the outlet flue and the wastewater from the liquid collection device, as well as the concentration of chemical oxygen demand in the wastewater from the liquid collection device, were collected multiple times.

[0030] According to the preset mapping rules, the combination of trend phenomena represented by the collected data is obtained. The preset mapping rules include: a. The chemical oxygen demand concentration in the wastewater of the liquid collection device changes with time within a preset time period, which is divided into slow increase, sudden increase or first decrease and then increase.

[0031] b. The trend of heavy metal concentration in the flue gas purified from the outlet flue over a preset period of time is classified as temporarily stable, slightly increasing, or significantly increasing.

[0032] c. The trend of heavy metal concentration in the wastewater of the collection device over a preset time period is classified as either decreasing or rapidly increasing.

[0033] d. Combine the changing trends in a, b, and c into trend phenomena.

[0034] It should be noted that the above-mentioned process for determining the trend of chemical oxygen demand concentration over time is as follows: multiple sets of chemical oxygen demand concentration values ​​are extracted within a preset time period using an electrochemical COD analyzer, the rate of change of chemical oxygen demand concentration at each sampling time point is quantified, and the linear regression slope of the overall time series is calculated. If the linear regression slope is positive, it indicates that the overall trend is upward.

[0035] Calculate the average rate of change per unit sampling time and the proportion of sampling points within the rising period. If the average rate of change per unit sampling time is less than or equal to a preset slow rate threshold, and the proportion of sampling points within the rising period is greater than a preset percentage, the preset percentage can be exemplarily 70%, then the trend of chemical oxygen demand concentration changing with time is determined to be a slow increase, where the rising period refers to a continuous period consisting of sampling time points where the rate of change of chemical oxygen demand concentration is greater than 0.

[0036] The sampling time series is divided into short time series. If the average rate of change per unit sampling time within a short time series is greater than or equal to a preset multiple of the preset slow rate threshold, and the absolute increase in the chemical oxygen demand concentration of the short time series is greater than or equal to the preset relative increase, then the trend of chemical oxygen demand concentration changing with time is determined to be a sudden increase.

[0037] The sampling time point with the minimum chemical oxygen demand (COD) concentration within the sampling time series is recorded as the reference time point. The sampling time series is then divided with the reference time point as the boundary. If the linear regression slope of the sampling time series segment before the reference time point is negative and the linear regression slope of the sampling time series segment after the reference time point is positive, then the reference time point is recorded as the inflection point. If there is an inflection point and the average rate of change per unit sampling time point after the inflection point is less than or equal to the preset slow rate threshold, then the trend of COD concentration change over time is determined to be first decreasing and then increasing.

[0038] The above-mentioned process for determining the trend of heavy metal concentration in wastewater over time is as follows: multiple sets of heavy metal concentration values ​​in wastewater are extracted by inductively coupled plasma mass spectrometry within a preset time period, and the linear regression slope of the overall time series is calculated. If the linear regression slope is negative, the trend of heavy metal concentration in wastewater over time is determined to be decreasing.

[0039] Consistent with the above method for determining whether the chemical oxygen demand concentration changes sharply over time, this method is used to determine whether the heavy metal concentration in wastewater changes sharply over time and whether the heavy metal concentration in flue gas subsequently changes sharply over time.

[0040] The above-mentioned process for determining the trend of heavy metal concentration in purified flue gas over time is as follows: multiple sets of heavy metal concentration values ​​in purified flue gas are extracted within a preset time period using laser-induced breakdown spectroscopy, and the standard deviation of the multiple sets of heavy metal concentration values ​​in purified flue gas within the preset time period is calculated. The specific calculation content is existing technology and will not be elaborated here.

[0041] If the standard deviation is less than or equal to the preset stable standard deviation threshold, and the deviation of the average concentration of heavy metals in the flue gas within the sampling period from the preset qualified limit of heavy metal concentration is less than or equal to the preset permissible deviation, then the trend of the change of heavy metal concentration in the flue gas over time is determined to be temporarily stable. The deviation calculation process is as follows: calculate the average concentration of heavy metals in the flue gas within the preset time period and calculate the difference with the preset qualified metal concentration threshold. Compare the difference with the preset qualified metal concentration threshold to obtain the deviation.

[0042] Consistent with the above method for determining that the chemical oxygen demand concentration changes slowly over time, this method is used to determine that the heavy metal concentration in flue gas changes slightly over time.

[0043] Among them, the preset slow rate threshold, preset relative increase, and preset stable compliance standard deviation threshold are all derived from the analysis of historical blockage level events in the electrode treatment. The preset slow rate threshold refers to the maximum value of the chemical oxygen demand concentration when it is lower than the low blockage level. The preset relative increase refers to the maximum increase of the chemical oxygen demand concentration when it is in the medium blockage level. The preset stable compliance standard deviation threshold refers to the maximum value of the standard deviation of the heavy metal concentration in the purified flue gas within a preset time period. All of the above are dimensionless and their numerical values ​​are used for logical analysis. The preset percentage refers to the ratio of the data where the change rate of chemical oxygen demand concentration and the change rate of heavy metal concentration in flue gas are both greater than 0 when the blockage level is low to the total change rate within the preset time period.

[0044] Reference Figure 3 As shown, in a preferred embodiment of the present invention, the process of constructing the pre-trained machine learning model includes:

[0045] (1) Collect historical operation data samples, the samples include multiple combinations of pollution parameter trends and corresponding actual plate blockage level labels, the level labels include low, medium and high.

[0046] (2) Divide the combination of multiple pollution parameter trend phenomena and corresponding blockage level labels into training set and validation set.

[0047] (3) The trend phenomenon combination is used as the input of the model, and the congestion level label is used as the output of the model.

[0048] By using the training set and a supervised learning mechanism, we capture various trend phenomenon combinations and their corresponding level labels to obtain a trained model. In the trained model, we train the mapping relationship between trend phenomenon combinations and their corresponding level labels.

[0049] (4) Detect the accuracy of congestion level prediction through the validation set, adjust the model hyperparameters, and supplement the model with similar historical running data samples for level labels with low prediction accuracy to perform incremental training until the accuracy reaches the preset threshold.

[0050] It should be noted that the blockage label types corresponding to the trend phenomenon combination include: when three of the following conditions are met simultaneously, the chemical oxygen demand concentration changes slowly or first decreases and then increases over time, the wastewater heavy metal concentration changes decreasing over time, and the flue gas heavy metal concentration changes temporarily stable or slightly increases over time, the output is a low blockage level label.

[0051] A slow increase in chemical oxygen demand (COD) concentration over time indicates that the plates still have a certain treatment capacity and the operating status is relatively stable. It is unlikely to cause serious clogging problems due to a sudden and large increase in organic pollutants. A decrease followed by an increase may be due to minor fluctuations in the influent water quality or slight changes in the system operating conditions. As long as the increase does not exceed the preset slow rate threshold, the plates can still maintain normal operation within a certain range. Therefore, both can be considered as the basis for corresponding low clogging levels.

[0052] The wastewater heavy metal concentration decreased over time, indicating that the electrode plate effectively removed heavy metals. The reduction in heavy metal concentration reduced the possibility of blockage caused by heavy metal precipitation or reaction with other substances, thus lowering the overall risk of blockage of the electrode plate. Therefore, this can be considered an important basis for determining a low blockage level.

[0053] When the concentration of heavy metals in the flue gas remains temporarily stable over time, it indicates that the electrode plates are operating relatively smoothly and have a continuous removal effect on heavy metals. The risk of electrode plate blockage is at a low level. A slight increase indicates that the electrode plates still have a certain processing capacity. This slight increase is within the tolerance range of the electrode plates and will not immediately lead to electrode plate blockage or affect the normal operation of the system. Therefore, it can still be used as a basis for determining a low blockage level.

[0054] In summary, when the above three conditions are met simultaneously, the electrode plate is in a relatively stable operating state, and the risk of blockage is low, so a low blockage level label is output.

[0055] When all three conditions are met simultaneously—a sudden increase in chemical oxygen demand (COD) concentration over time, a sharp increase in wastewater heavy metal concentration over time, and a significant increase in flue gas heavy metal concentration over time—the output will be a high clogging level label.

[0056] The chemical oxygen demand (COD) concentration increases abruptly over time, causing the plate load to increase instantaneously, exceeding the plate's normal processing capacity. Undecomposed organic matter may accumulate in the plate pores, forming viscous substances that gradually clog the pores, affecting the normal flow of wastewater and the treatment effect, and increasing the likelihood of plate blockage.

[0057] The concentration of heavy metals in wastewater increases sharply over time. Heavy metal ions readily react with other substances in the water to form insoluble precipitates. These precipitates accumulate in the pores of the electrode plates, gradually clogging them and greatly increasing the likelihood of plate blockage.

[0058] The flue gas heavy metal concentration showed a significant upward trend over time, indicating that the electrode plate had a very poor effect on treating the flue gas and could not operate normally, reflecting severe blockage of the electrode plate pores.

[0059] In summary, when all three conditions are met simultaneously, the electrode plate cannot operate normally, the electrode plate holes are severely blocked, and therefore a high blockage level label is output.

[0060] Other trend phenomena combinations are output as medium congestion level labels.

[0061] In a preferred embodiment of the present invention, the determination of the distribution pattern of electrode blockage is based on the following: the electrode channel is divided into multiple partition channels, and differential pressure transmitters are installed in the partition channels to obtain the pressure drop data of each partition channel.

[0062] It should be noted that a differential pressure transmitter is installed in each zone channel, with one end of the differential pressure transmitter close to the smoke inlet and the other end close to the smoke outlet. The pressure drop data of the corresponding zone channel is obtained by displaying the value of the differential pressure transmitter. This arrangement makes the monitored data more consistent with reality.

[0063] The pressure drop data between multiple adjacent spraying time points in each zone channel are statistically analyzed to calculate the overall variance of the electrode channel. The overall variance is then compared with a first preset variance and a second preset variance, wherein the first preset variance is greater than the second preset variance.

[0064] Calculate the linear correlation coefficient of pressure drop data relative to time between multiple adjacent spraying time points in each zone channel, determine whether the degree of blockage in each zone channel is linearly or non-linearly related to the operating time, and thus determine the trend relationship between the degree of blockage in the plate channel and the operating time.

[0065] It should be noted that the aforementioned linear correlation coefficient specifically refers to the Pearson correlation coefficient, which can be calculated based on the standard deviation of the pressure drop series composed of pressure drop data between multiple adjacent spraying time points, the standard value of the corresponding time series, and the covariance of the two. The specific calculation content is existing technology and will not be elaborated here.

[0066] Judgment Criteria: The absolute value of the linear correlation coefficient between the pressure drop data and time at multiple adjacent spraying time points in a certain zone channel is compared with a preset linear compliance threshold. If it is greater than or equal to the threshold, it is determined that the degree of blockage in the zone channel is linear with the operating time. By collecting multiple sets of experimental data with a known linear relationship between pressure drop and time, the correlation coefficients of the multiple sets of linear data are calculated, thereby obtaining a series of values ​​reflecting the strength of the linear relationship between pressure drop and time. The average value of the above series of values ​​is calculated, and the value obtained by the average value calculation is used as the preset linear compliance threshold. The preset linear compliance threshold can be 0.7 for example.

[0067] It should be noted that the process of determining the trend relationship between the degree of blockage of the plate channel and the running time is as follows: statistical data on the linear and non-linear correlation between the degree of blockage of each partition channel and the running time are collected, and the linear and non-linear quantities are compared. If the linear quantity is greater than or equal to the non-linear quantity, it is determined that the degree of blockage of the plate channel and the running time are linear. If the linear quantity is less than the non-linear quantity, it is determined that the degree of blockage of the plate channel and the running time are non-linear. The reason for determining that the degree of blockage of the plate channel and the running time are linear when the linear quantity is equal to the non-linear quantity is that the system may fluctuate due to environmental interference during processing. Using linearity has a stronger fault tolerance for data fluctuations and a lower risk of overfitting.

[0068] By statistically analyzing the pressure drop datasets collected at adjacent spraying time points, the pressure drop change rate before and after multiple spraying time points in each zone channel was calculated, and the pressure drop change rate of the electrode channel was obtained by double averaging.

[0069] It should be noted that the rate of change of pressure drop before and after the spraying time point is obtained by comparing the pressure drop after spraying with the pressure drop before spraying, and then comparing the pressure drop difference with the pressure drop before spraying.

[0070] When the plate channel simultaneously satisfies the following conditions: the overall variance is less than or equal to the second preset variance, the rate of change of pressure drop before and after spraying is less than or equal to the preset rate of change of pressure drop threshold, and the trend relationship between the degree of blockage and the running time is linear, the plate blockage condition is identified as instantaneous blockage.

[0071] When the overall variance is greater than the first preset variance, the rate of change of pressure drop before and after spraying is greater than the preset threshold for the rate of change of pressure drop, and the trend relationship between the degree of blockage and the running time is non-linear, it is identified as persistent blockage.

[0072] When the overall variance is within the range defined by the first and second preset variances, it is identified as complex blockage.

[0073] In a preferred embodiment of the present invention, the main causes of the plate blockage distribution pattern are traced as follows:

[0074] If the blockage is persistent, the main cause is insufficient spray pressure; if the blockage is transient, the main cause is insufficient spray frequency; and if the blockage is a combination of both, the main cause is insufficient spray pressure and insufficient spray frequency.

[0075] It should be noted that the preset pressure drop change rate threshold refers to the maximum critical point for the degree of improvement in blockage before and after spraying. The maximum critical point is determined through experiments. Specifically, it is the pressure drop change rate fed back after a single spraying operation on a batch of electrode partition channels with different degrees of blockage, where the electrode partition channels are in a blockage-free state. The maximum value of this value is selected as the preset pressure drop change rate threshold.

[0076] The first preset variance is the minimum critical point characterizing the uneven distribution of blockage in the electrode channel. This minimum critical point is determined experimentally. Specifically, it involves selecting samples from historical data where the degree of blockage varies significantly across different zones, with severe blockage in localized areas (such as localized dust accumulation due to insufficient long-term spraying), and statistically analyzing the range of variance values ​​for these samples. The minimum value within this range is taken as the first preset variance. That is, when the overall variance is greater than or equal to the first preset variance, it can be determined that the blockage status of each zone of the electrode is in a relatively uneven range.

[0077] The second preset variance is the maximum critical point of variance that characterizes the relatively uniform blockage situation in the channel of the characterizing plate. The maximum critical point of variance is determined by experiment. Specifically, experimental samples with no significant difference in the degree of blockage in each zone and uniform overall distribution (such as the state when the equipment is just started or the spraying effect is ideal) are prepared, and the range of variance values ​​of the sample set is statistically analyzed. The maximum value in this range is taken as the second preset variance. That is, when the overall variance is less than or equal to the second preset variance, it can be determined that the blockage state of each zone of the plate is in a relatively uniform range.

[0078] It should also be noted that when the overall variance of the electrode channel is less than or equal to the second preset variance, it indicates that the pressure drop data is relatively concentrated, the operating state of the electrode channel is relatively stable, the rate of change of pressure drop before and after spraying is less than or equal to the preset threshold for the rate of change of pressure drop, and the trend relationship between the degree of blockage and the operating time is linear. Therefore, it is judged as transient blockage. When the interval between spraying is too long, a single spray can clear the blockage. It can be eliminated by adjusting the spraying frequency. Therefore, the main cause of transient blockage is insufficient spraying frequency.

[0079] When the overall variance of the electrode channel is greater than the first preset variance, it indicates that the pressure drop data is relatively dispersed, the operating state of the electrode channel is unstable, the rate of change of pressure drop before and after spraying is greater than the preset threshold for the rate of change of pressure drop, and the trend relationship between the degree of blockage and the operating time is non-linear, so it is judged as continuous blockage. When a single spray cannot completely clear the blockage and there is still a large area of ​​blockage, it is necessary to adjust the spray pressure to clear the blockage. Therefore, the main cause of continuous blockage is insufficient spray pressure.

[0080] This invention achieves accurate clogging level determination by mapping the combination of pollution parameter trends with clogging levels through machine learning. By tracing the main causes of clogging through the distribution pattern of plate clogging, it breaks through the limitation of traditional fixed modes that cannot accurately understand the real-time clogging situation of equipment. This accurate determination and tracing provides an accurate basis for subsequent adjustment of spraying strategies.

[0081] The spray adjustment module dynamically adjusts the spray parameters according to the plate blockage level and the main cause of blockage to perform spray adjustment. The spray parameters include spray frequency and spray pressure.

[0082] In a preferred embodiment of the present invention, the dynamic adjustment process of the spray parameters includes triggering spray adjustment when the machine learning model outputs a clogging or high-clogging level label, and executing the following dynamic adjustment strategy according to the main cause of clogging:

[0083] When the cause of the blockage is found to be insufficient spray pressure or insufficient spray frequency, the corresponding spray parameter adjustment value can be retrieved by consulting the preset mapping table according to the blockage level label.

[0084] When the cause of the blockage is found to be insufficient spray pressure and spray frequency, the preset mapping table is queried according to the blockage level label, the adjustment amplitude of the corresponding spray parameters is retrieved, and the influence weight of spray pressure and spray frequency is allocated according to the pressure drop change rate before and after spraying to adjust the adjustment amplitude accordingly.

[0085] It should be noted that the formula for calculating pressure weight is: The frequency weight can be obtained by the difference between 1 and the pressure weight.

[0086] In the formula: The pressure drop change rate before and after spraying is given by k, which is a preset slope factor (k can be set to 0.2 by default). To preset the threshold for the rate of change of pressure drop, As a preset weighting factor,

[0087] The preset weighting factor is the minimum baseline influence weight representing the spray pressure under complex blockage conditions. The baseline influence weight can be obtained by relevant personnel based on their professional experience through historical complex blockage adjustment operation data.

[0088] In a preferred embodiment of the present invention, the preset mapping table is composed of spray pressure amplitude and spray frequency amplitude corresponding to different blockage level labels.

[0089] It should be noted that the spray pressure amplitude and spray frequency amplitude corresponding to the preset mapping table are calculated based on the required flushing force based on the pipe size and fluid characteristics, taking into account the properties of the blockage (hardness, viscosity, etc.) to determine the required impact force, and calculating the minimum energy input required to clear different degrees of blockage.

[0090] According to the blockage level and cause, the embodiments of the present invention quickly adjust the spray parameters by combining a dynamic spray adjustment strategy, avoiding the lag of existing manual trial and error, improving the response rate of the equipment under various conditions, realizing on-demand cleaning, effectively solving the problem of water waste and insufficient cleaning effect, thereby improving the overall purification efficiency and operating economy of the equipment.

[0091] The execution feedback module analyzes the spray performance within a preset period after spray adjustment. If the performance meets the processing standard, the current spray parameters are automatically saved as the optimization benchmark value. If the performance does not meet the standard, the spray parameters are adjusted according to the gradient until they meet the processing standard.

[0092] In a preferred embodiment of the present invention, the spraying performance is as follows: the plate blockage level within a preset period after spraying adjustment is obtained, and it is determined whether the blockage level has reached a low level.

[0093] After monitoring and adjusting, at the end of the preset cycle, the concentrations of heavy metals in the flue gas purified by the flue gas outlet, the concentrations of heavy metals in the wastewater from the liquid collection device, and the concentrations of chemical oxygen demand are analyzed to determine whether their concentrations have decreased to the preset qualified limits.

[0094] The pressure drop data at the end of the preset cycle for each zone channel is statistically analyzed to calculate the pressure drop characterization coefficient, and it is then determined whether the pressure drop characterization coefficient has dropped to the preset low-level pressure drop coefficient threshold.

[0095] If the following conditions are met simultaneously: the heavy metal concentration in the flue gas purified from the outlet flue, the heavy metal concentration and chemical oxygen demand concentration in the wastewater from the liquid collection device decrease to the preset qualified limits, and the pressure drop characterization coefficient decreases to the preset low-level pressure drop coefficient threshold, then the treatment standard is met; otherwise, a gradient adjustment is performed.

[0096] It should be noted that the above-mentioned preset compliance limits are parameters that are modified by referring to industry standard emission requirements and taking into account flue gas volume, emission height and sensitivity to the surrounding environment. Specifically, they refer to the highest values ​​that can be emitted under low clogging levels for heavy metal concentration and chemical oxygen demand concentration.

[0097] The preset low-level pressure drop coefficient threshold is obtained by analyzing the pressure drop variation law under different loads and medium characteristics in combination with the system design parameters. The preset low-level pressure drop coefficient threshold refers to the maximum pressure drop variation rate of the system under low blockage level.

[0098] In a preferred embodiment of the present invention, the voltage drop characterization coefficient is formed by performing a ratio analysis on the voltage drop data at the end of a preset cycle of each partition channel and a preset reference voltage drop data, and then calculating the average of the ratio analysis results to obtain the voltage drop characterization coefficient of the plate channel.

[0099] In a preferred embodiment of the present invention, the gradient adjustment includes:

[0100] When the concentrations of heavy metals in the flue gas purified by the outlet flue, the heavy metal concentrations in the wastewater of the liquid collection device, and the chemical oxygen demand concentration all decrease to the corresponding preset qualified limits, and the pressure drop characterization coefficient does not decrease to the preset low-level pressure drop coefficient threshold, the spray parameters are adjusted according to the preset first gradient corresponding adjustment amplitude.

[0101] When any two of the above conditions are met, the spray parameters are adjusted according to the preset second gradient adjustment amplitude. When one of the above conditions is met, the spray parameters are adjusted according to the preset third gradient adjustment. The adjustment amplitudes are sorted in descending order as the preset first gradient, the preset second gradient, and the preset third gradient.

[0102] It should be noted that the aforementioned preset gradient adjustment amplitude is obtained by referring to the spray pressure amplitude and spray frequency amplitude in the preset mapping table, and taking a preset percentage value based on these values. The preset percentage is obtained by collecting historical adjustment event data, recording key information of each adjustment event, including the total adjustment amplitude and the number of adjustments, filtering out invalid events of over-adjustment, conducting single adjustment error experimental analysis on valid adjustment events, and retrieving the percentage range that can balance the adjustment speed and avoid oscillations. The median value of the range is used as the preset percentage. The preset first gradient can be exemplarily 15%. For example, when the corresponding spray pressure amplitude is 1.2, the gradient adjustment pressure amplitude is 0.18.

[0103] This invention uses the spraying effect to indicate whether the spraying adjustment is appropriate, and then uses gradient adjustment to quickly iterate to the optimal combination of spraying parameters. This iterative optimization mechanism based on operational feedback and flexible amplitude adjustment avoids the problems of excessive investment and ineffective consumption caused by fixed amplitude adjustment, reduces operating costs, and achieves closed-loop automated operation by continuously providing feedback and adjustment based on the spraying effect.

[0104] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0106] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0108] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0109] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-temperature flue gas treatment intelligent control system based on a wet electrostatic precipitator, characterized in that, include: The information acquisition module monitors the pollution parameters of the flue gas purified by the outlet flue of the wet electrostatic precipitator and the wastewater of the liquid collection device. The pollution parameters include the concentration of heavy metal pollutants and the concentration of chemical oxygen demand. The data processing module performs trend feature analysis on the pollution parameters to obtain trend phenomenon combinations, inputs them into a pre-trained machine learning model to determine the plate blockage level, and traces the main causes of blockage based on the plate blockage distribution pattern. The spray adjustment module dynamically adjusts the spray parameters according to the plate blockage level and the main cause of blockage to perform spray adjustment. The spray parameters include spray frequency and spray pressure. The execution feedback module analyzes the spray performance within a preset period after spray adjustment. If the spray performance meets the processing standard, the current spray parameters are automatically saved as the optimization benchmark value. If they do not meet the standard, the spray parameters are adjusted according to the gradient until they meet the processing standard. The specific processing procedure for trend characteristic analysis of the pollution parameters The process includes the following steps: collecting heavy metal concentrations in the purified flue gas from the outlet flue and the wastewater from the liquid collection device, as well as chemical oxygen demand concentrations in the wastewater from the liquid collection device, multiple times within a preset time period. The trend phenomena represented by the collected data are obtained according to preset mapping rules. The preset mapping rules include: a) the trend of chemical oxygen demand (COD) concentration in the wastewater of the liquid collection device over a preset time period, which is divided into slow increase, sudden increase, or decrease followed by increase; b) the trend of heavy metal concentration in the flue gas from the outlet flue over a preset time period, which is divided into temporary stability, slight increase, or significant increase; c) the trend of heavy metal concentration in the wastewater of the liquid collection device over a preset time period, which is divided into decrease or sharp increase; d) combining the trends in a, b, and c into trend phenomena. The process of constructing the pre-trained machine learning model includes: (1) collecting historical running data samples, the samples including multiple combinations of pollution parameter trend phenomena and corresponding actual plate blockage level labels, the level labels including low, medium and high; (2) dividing multiple combinations of pollution parameter trend phenomena and corresponding blockage level labels into training set and validation set; (3) using the trend phenomenon combination as the input of the model and the blockage level label as the output of the model; using the training set through a supervised learning mechanism, capturing each trend phenomenon combination and its corresponding level label to obtain a trained model, and training the mapping relationship between the trend phenomenon combination and its corresponding level label in the trained model; (4) detecting the accuracy of blockage level prediction through the validation set, adjusting the model hyperparameters, and adjusting the model for those with low prediction accuracy. Level labels are added to supplement historical operational data samples of the same type for incremental training of the model until the accuracy reaches a preset threshold. The blockage label types corresponding to the trend phenomenon combinations are as follows: when three of the following conditions are met simultaneously, the output is a low blockage level label: the chemical oxygen demand concentration changes slowly or first decreases and then increases over time, the wastewater heavy metal concentration changes decreasing over time, and the flue gas heavy metal concentration changes temporarily stable or slightly increases over time. When three of the following conditions are met simultaneously, the output is a high blockage level label: the chemical oxygen demand concentration changes suddenly increases over time, the wastewater heavy metal concentration changes sharply increases over time, and the flue gas heavy metal concentration changes significantly increases over time. Other trend phenomenon combinations are output as medium blockage level labels. The determination of the electrode plate blockage distribution pattern is based on the following: The electrode plate channel is divided into multiple partition channels, and differential pressure transmitters are installed in the partition channels to obtain pressure drop data for each partition channel; the pressure drop data between multiple adjacent spraying time points in each partition channel are statistically analyzed to calculate the overall variance of the electrode plate channel, and the overall variance is compared with a first preset variance and a second preset variance, wherein the first preset variance is greater than the second preset variance; the linear correlation coefficient between the pressure drop data between multiple adjacent spraying time points in each partition channel and time is calculated to determine whether the blockage degree of each partition channel is linearly or non-linearly correlated with the operating time, thereby determining the trend relationship between the blockage degree of the electrode plate channel and the operating time; statistical analysis of each... The pressure drop change rate before and after multiple spraying time points in each zone channel was calculated using double averaging to obtain the pressure drop change rate of the electrode channel. When the electrode channel simultaneously satisfies the following conditions: the overall variance is less than or equal to the second preset variance, the pressure drop change rate before and after spraying is less than or equal to the preset pressure drop change rate threshold, and the trend relationship between the degree of blockage and the running time is linear, the electrode blockage is identified as transient blockage. When the overall variance is greater than the first preset variance, the pressure drop change rate before and after spraying is greater than the preset pressure drop change rate threshold, and the trend relationship between the degree of blockage and the running time is non-linear, it is identified as persistent blockage. When the overall variance is within the interval defined by the first and second preset variances, it is identified as complex blockage. The main causes of the plate blockage distribution pattern can be traced as follows: if the blockage is continuous, the main cause is insufficient spray pressure; if the blockage is momentary, the main cause is insufficient spray frequency; and if the blockage is complex, the main cause is insufficient spray pressure and insufficient spray frequency.

2. The intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator according to claim 1, characterized in that: The dynamic adjustment process of the spray parameters includes triggering spray adjustment when the machine learning model outputs a clogging or high-clogging level label, and executing the following dynamic adjustment strategy based on the main cause of clogging: When the cause of the blockage is found to be insufficient spray pressure or insufficient spray frequency, the corresponding spray parameter adjustment value can be retrieved by consulting the preset mapping table according to the blockage level label. When the cause of the blockage is found to be insufficient spray pressure and spray frequency, the preset mapping table is queried according to the blockage level label, the adjustment amplitude of the corresponding spray parameters is retrieved, and the influence weight of spray pressure and spray frequency is allocated according to the pressure drop change rate before and after spraying to adjust the adjustment amplitude accordingly.

3. The intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator according to claim 2, characterized in that: The preset mapping table consists of spray pressure amplitude and spray frequency amplitude corresponding to different blockage level labels.

4. The intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator according to claim 1, characterized in that: The spraying performance is as follows: the plate blockage level within a preset cycle after spraying adjustment is obtained, and it is determined whether the blockage level has reached a low level. After monitoring and adjusting, at the end of the preset cycle, the concentration of heavy metals in the flue gas purified by the flue outlet, the concentration of heavy metals in the wastewater of the liquid collection device, and the concentration of chemical oxygen demand are analyzed to determine whether their concentrations have decreased to the corresponding preset qualified limits. Statistical analysis of pressure drop data at the end of the preset cycle time of each zone channel is used to calculate the pressure drop characterization coefficient, and it is determined whether the pressure drop characterization coefficient has dropped to the preset low-level pressure drop coefficient threshold. If the following conditions are met simultaneously: the heavy metal concentration in the flue gas purified from the outlet flue, the heavy metal concentration in the wastewater from the liquid collection device, and the chemical oxygen demand concentration decrease to the corresponding preset qualified limits, and the pressure drop characterization coefficient decreases to the preset low-level pressure drop coefficient threshold, then the treatment standard is met; otherwise, a gradient adjustment is performed.

5. The intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator according to claim 4, characterized in that: The voltage drop characterization coefficient is composed of a ratio analysis of the voltage drop data at the end of a preset cycle in each partition channel and a preset benchmark voltage drop data, and the average value of the ratio analysis results is calculated to obtain the voltage drop characterization coefficient of the plate channel.

6. The intelligent control system for high-temperature flue gas treatment based on a wet electrostatic precipitator according to claim 4, characterized in that: The gradient adjustment includes: When the following conditions are met simultaneously: the blockage level has not decreased to the low-level label, the heavy metal concentration of the flue gas purified at the outlet flue, the heavy metal concentration and chemical oxygen demand concentration of the wastewater from the liquid collection device have decreased to the corresponding preset qualified limits, and the pressure drop characterization coefficient has not decreased to the preset low-level pressure drop coefficient threshold, the spray parameters are adjusted according to the preset first gradient corresponding adjustment amplitude. When any two of the above conditions are met, the spray parameters are adjusted according to the preset second gradient adjustment amplitude. When any one of the above conditions is met, the spray parameters are adjusted according to the preset third gradient adjustment. The adjustment amplitudes are sorted in descending order as the preset first gradient, the preset second gradient, and the preset third gradient.

Images