A method for zoned airflow control in jet valve cleaning

By establishing a mapping between zones and jet valves and resistance baseline data, eliminating global fluctuations, and generating directional airflow transfer data, the problem of inaccurate airflow prediction in zoned dust removal control is solved, and safe and reliable zoned airflow control is achieved.

CN122828481BActive Publication Date: 2026-11-10CHANGZHOU KEKANG ENVIRONMENTAL PROTECTION DUST REMOVAL EQUIP CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611358937.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-09-03
Publication Date
2026-11-10
Estimated Expiration
2046-09-03

AI Technical Summary

Technical Problem

Existing zone cleaning control methods are unable to accurately predict the residual effects of cross-zone directional airflow transfer and global operating condition fluctuations caused by local jet cleaning, leading to problems such as zone airflow exceeding limits or mutual interference between adjacent jet cleaning processes.

Method used

By acquiring the total air volume and operating parameters of each zone, a mapping between the zone and the jet valve is established, resistance baseline data is generated, global fluctuations during non-jet periods are eliminated, the target area is determined, and directional air volume transfer data is generated based on the jet event response data. The resistance baseline data is updated, the air volume of each zone is predicted, zones exceeding limits are excluded, and the dust removal triggering conditions are determined.

Benefits of technology

It reduces the interference of operating load changes on target area judgment and event response, accurately predicts the residual impact of historical injection and air volume recovery, ensures that the air volume of the zone after injection is within a safe range, and avoids air volume exceeding the limit and interference from adjacent injections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122828481B_ABST
    Figure CN122828481B_ABST
Patent Text Reader

Abstract

This application provides a method for zoned airflow control in pulse-jet valve cleaning, relating to the field of baghouse dust collector cleaning control. The method involves acquiring the total airflow and operating parameters of each zone, eliminating global fluctuations during non-pulse-jet periods to determine the target area; controlling the pulse-jet valve to perform the target pulse-jet, eliminating global fluctuations before pulse-jet and generating pulse-jet event response data; generating directional airflow transfer data based on the response change direction parameters, response change amplitude parameters, response delay parameters, and response attenuation parameters of non-target areas, and updating the resistance baseline data based on the response of the target area; predicting the airflow of each zone by combining the current residual coupling amount, the evaluated residual coupling amount, the post-pulse airflow recovery curve, and the predicted transfer amount, eliminating candidate zones that would cause the predicted airflow to exceed the limit, determining and verifying the next target area, and continuing or ending cleaning after verification. This application can reduce the risk of zoned airflow exceeding limits and mutual interference during continuous pulse-jet cleaning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of dust collector cleaning control technology, and in particular to a zoned airflow control method for cleaning pulse-jet valves. Background Technology

[0002] Baghouse dust collectors typically divide the filter bag assembly into multiple zones, with one or more pulse-jet valves configured for each zone. These valves release compressed gas according to cleaning control commands, causing a transient reverse airflow or deformation in the corresponding filter bag to remove accumulated dust from its surface. The zone's outlet airflow and filtration pressure differential reflect changes in filter bag resistance and are commonly used operating parameters for determining the target material and timing of cleaning.

[0003] Existing zone cleaning control typically selects the target for pulse-jet cleaning based on the current zone's output air volume, filtration differential pressure, or a preset sequence. When multiple zones share a fan and part of the ductwork, pulse-jet cleaning of one zone will change its filtration resistance and redistribute airflow to other zones; this redistribution has differences in direction, amplitude, delay, and attenuation. Simultaneously, changes in fan operating points and production loads will cause synchronous changes in each zone, resulting in the pulse-jet cleaning effect being mixed with global operating condition fluctuations.

[0004] Therefore, due to the combined effects of cross-zone directional airflow transfer caused by localized jet cleaning and global operating condition fluctuations, the existing control process has difficulty in determining the residual effects of previous jet cleaning that will still exist in the next jet cleaning, and it is also difficult to predict the airflow of each zone after the residual effects are superimposed with the response of the next jet cleaning. During continuous dust cleaning, problems such as zone airflow exceeding the limit or mutual interference between adjacent jet cleaning are likely to occur. Summary of the Invention

[0005] This application provides a zoned airflow control method for jet valve cleaning, which addresses the problem that the residual effects of previous jet cleaning are difficult to determine and the airflow of each zone after the next jet cleaning is difficult to accurately predict due to the combined effects of cross-zone directional airflow transfer caused by local jet cleaning and global operating condition fluctuations. To achieve the above objective, this application adopts the following technical solution: A method for controlling the airflow in a zone for cleaning a jet valve, the method comprising: The system acquires the total air volume and operating parameters of each zone, establishes a mapping between zones and jet valves, and generates resistance baseline data. It then removes global fluctuations during non-jet periods from the operating parameters of each zone to generate zone deviation data, and determines the target area based on this data. The system controls the jet valves corresponding to the target area to perform target jetting according to the mapping, removing global fluctuations before jetting and generating jetting event response data. It generates directional air volume transfer data from the target area to non-target areas based on the jetting event response data. The system updates the resistance baseline data based on the target area and the jetting event response data. It predicts the air volume of each zone by combining the directional air volume transfer data and the resistance baseline data, excludes zones with predicted air volume exceeding limits, designates the remaining zones as candidate zones, and determines the next target area from the candidate zones. It determines whether the next target area meets the dust removal triggering conditions. If it does, the next target area is set as the new target area, and the system returns to the target jetting execution step. If it does not meet the conditions, the dust removal process ends.

[0006] As can be seen from the above technical solution, this application has the following beneficial effects: 1. This application removes the corresponding global fluctuation components from the zonal operating parameters during non-purge periods and the operating parameters during the purge response period, so that the zonal deviation data can be used to characterize local resistance anomalies, and the purge event response data can be used to characterize the local and cross-zonal responses caused by a single purge, thereby reducing the interference of operating load changes on target area judgment and event response extraction.

[0007] 2. This application saves the response change direction parameters, response change amplitude parameters, response delay parameters, and response attenuation parameters according to the directional relationship from the target area to each non-target area, and calculates the current residual coupling amount at the time to be executed and the evaluation residual coupling amount at the time of prediction and evaluation respectively; thereby, before executing the next injection, it can predict the superposition result of the residual impact of historical injection, the recovery of air volume in the target area, and the predicted transfer of non-target areas, and exclude candidate areas that would cause the predicted air volume of any zone to exceed the preset safety limit.

[0008] 3. This application updates the resistance baseline data based on the measured recovery trajectory of the target area, and re-verifies the dust removal triggering conditions before the next target area is purged; after purging, the recovery curve and directional air volume transfer data are corrected by the deviation between the actual response value and the predicted response value, so that subsequent predictions use data that are compatible with the current filter bag resistance state and zone coupling state. Attached Figure Description

[0009] The present application will be further described below with reference to the accompanying drawings.

[0010] Figure 1 A flowchart illustrating the zonal airflow control method provided in this application embodiment; Figure 2This is a flowchart illustrating the generation of directional airflow transfer data and the calculation of residual coupling amount provided in the embodiments of this application. Figure 3 A flowchart for candidate zone airflow prediction and screening provided in this application embodiment. Detailed Implementation

[0011] The terms "first," "second," and "third," etc., used in this application specification, claims, and drawings are used to distinguish different objects, not to limit a specific order.

[0012] Research has revealed that in baghouse dust collectors where multiple zones share a fan and part of the ductwork, a single localized pulse-jet cleaning not only alters the filtration resistance of the target area but also affects non-target areas with varying directions, amplitudes, delays, and attenuations. If the ranking is based solely on the measured airflow before the next pulse-jet, the measured values ​​still contain the residual airflow from previous pulse-jet cleanings and cannot reflect the additional impact on other zones after the next target area is cleaned. Therefore, this method cannot be directly used for safety assessments related to continuous pulse-jet cleaning.

[0013] To address the aforementioned issues, this application first separates global fluctuations from operational data and injection response data, then extracts the direction, amplitude, delay, and attenuation of a single injection according to directed partitions; before determining the next target area, it calculates the residual air volume impact of historical injections at the time of execution and the time of prediction and evaluation, and superimposes the target recovery and non-target transfer of candidate injections to obtain the predicted air volume for each partition.

[0014] Example: like Figures 1-3 As shown, in this embodiment, the filter bag groups of the bag filter are divided into multiple zones according to independent outlet branches or filter bag groups that can be independently pulsed cleaned. Each zone is assigned a zone identifier and a corresponding relationship is established with one or more pulsed cleaned valve identifiers. The total air volume is collected by the air volume sensor in the main outlet duct; the zone outlet air volume is collected by the air volume sensor in the outlet branch of each zone; the zone filtration pressure difference is collected by the differential pressure sensor connecting the dust-facing side and the clean air side of the filter bag respectively. The controller timestamps the data from each sensor, the pulsed cleaned command sending time, the pulsed cleaned valve opening time, and the pulsed cleaned valve closing time using a unified clock.

[0015] In this embodiment, the target area refers to the object of the current target pulse-jet action; the non-target area refers to the partitions in the same dust collector other than the target area. The candidate partitions are those that are successively substituted into the prediction model to evaluate their pulse-jet consequences; the candidate partitions are those in the candidate partitions whose predicted air volume for each partition in the entire group does not exceed their respective preset safety limits; the next target area is the candidate partition with the highest dust removal priority and is used for the next target pulse-jet after the dust removal trigger conditions have been verified.

[0016] The zonal operating parameters referred to in this embodiment include at least the zonal airflow and zonal filtration differential pressure. The unit of zonal airflow is cubic meters per second, and the unit of zonal filtration differential pressure is Pascals. Global fluctuations are variations caused by changes in the fan operating point, inlet airflow, or production load, and occur synchronously in multiple zones. The global fluctuation component is not a dimensionless composite value, but a data vector composed according to the type of operating parameter and the zonal index; the airflow component is subtracted only from the airflow, and the filtration differential pressure component is subtracted only from the filtration differential pressure.

[0017] The jetting event response data referred to in this embodiment is a time series formed by removing global fluctuations before jetting from the operating parameters of each zone during the jetting response period and then filtering by moving average. The directional airflow transfer data uses directed zone pairs from the target area to a non-target area as indices, storing data records of the non-target area's response direction, magnitude, delay, and attenuation parameters to the target jetting. Data from the target area to the first non-target area and data from the target area to the second non-target area are stored separately, without merging different directions into a directionless correlation coefficient.

[0018] The resistance baseline data referred to in this embodiment is a reference data structure based on zoning identifiers and total airflow operating condition interval indexes. It includes at least the baseline airflow, baseline filter differential pressure, post-purge airflow recovery curve, curve sampling interval, and update frequency. The baseline airflow represents the zoning airflow within the corresponding total airflow operating condition interval, formed by the initial non-purge calibration samples and subsequent samples that have passed zoning deviation verification, free from global fluctuations. The post-purge airflow recovery curve represents the recovery change of the target area airflow relative to the baseline value at the purge execution start point, with the purge execution start point as the zero point of time. This data structure characterizes the filter resistance state through both zoning airflow and filter differential pressure, without requiring the conversion of filter bag resistance into a single absolute resistance value.

[0019] In this embodiment, the positive or negative airflow change still caused by a single historical dust removal action to the affected zone at the time to be calculated is called the residual airflow impact. The current residual coupling amount is the sum of the residual airflow impact amounts corresponding to the same zone at the time to be executed, and the evaluation residual coupling amount is the sum of the residual airflow impact amounts corresponding to the same zone at the prediction and evaluation time. A positive value indicates an increase in the airflow of the corresponding zone, and a negative value indicates a decrease in the airflow of the corresponding zone; both are in cubic meters per second. The prediction benchmark value is the airflow benchmark obtained by subtracting the current residual coupling amount from the measured zone airflow at the time to be executed, and does not include the residual impact of historical jet cleaning.

[0020] This embodiment sequentially performs initial calibration, online dust removal control, and online data updates. The controller first uses non-jet calibration samples to form a resistance benchmark, airflow distribution coefficient, differential pressure sensitivity coefficient, and deviation threshold. Then, it deletes samples exceeding limits using stability and deviation thresholds and recalculates until the sample set no longer changes. Next, when the fan is running stably and there are no concurrent jets, it performs isolated jet cleaning zone by zone. The response from the target area forms the airflow recovery curve after jet cleaning, and the response from non-target areas forms the directional airflow transfer data for each directional zone pair. Automatic continuous dust removal is enabled after all calibration data is valid. During operation, after each effective jet cleaning, the recovery curve is updated with the measured recovery trajectory, and the directional airflow transfer data is corrected based on the deviation between the actual and predicted responses. If any zone lacks valid calibration data due to filter bag replacement, sensor failure, or data invalidation, automatic continuous dust removal is paused and recalibrated; this situation is not treated as an empty candidate list.

[0021] The sampling frequency is determined based on the bandwidth of the airflow sensor and the duration of the jet pulse, forming a continuous sampling sequence during the jet pulse that can distinguish the start and end of the pulse and changes in response. The synchronous sampling length before the start of the jet execution increases progressively according to the sampling period. The average rate of change of total airflow corresponding to adjacent candidate lengths is calculated, and the shortest candidate length is selected where the difference between adjacent calculation results does not exceed the rate resolution limit obtained by dividing the sensor's minimum resolvable change by the shorter candidate length. The first preset window is the upper empirical quantile of the elapsed time from the end of the jet pulse to the last time the response amplitude exceeds the response significance threshold in the valid calibration event. The preset evaluation interval is the upper empirical quantile of the response delay for each valid directed partition and aligned upwards to the sampling period. The second preset window is determined by the shortest time scale of the resolvable change in ash accumulation state and the solvability condition of the linear trend equation of total airflow. All the above parameters are written into the configuration record and include the version of the calibration data used.

[0022] The permissible misjudgment frequency or permissible omission frequency referred to in this embodiment is the ratio of the number of corresponding erroneous events allowed by the equipment safety procedures or maintenance strategies to the number of valid judgments, which is read from the controller configuration record. The upper empirical quantile is obtained according to a unified method: the valid historical samples are arranged in ascending order of value, the total number of samples is multiplied by the corresponding permissible frequency and rounded up to obtain the number of samples excluded from the top of the sequence, and then the maximum value among the remaining samples is used as the empirical quantile; if multiple samples have the same value, their original value is retained without interpolation. This method is used for threshold or window setting of steady state, partition deviation exceeding limit judgment, response significance, response delay, and residual amplitude, respectively, to ensure consistency in configuration source and calculation method.

[0023] Specific implementation steps: S1. Obtaining operating parameters, establishing partition mapping, and generating resistance benchmark data: The controller reads the total air volume, the air volume of each zone, and the filter pressure difference of each zone, and aligns them to a unified sampling time according to the timestamp. For data returned by the sensors between adjacent unified sampling times, linear interpolation is used to generate corresponding sampled values; when the continuous missing time exceeds the duration of one pulse, the current time period is marked as an invalid time period, and the target pulse is not initiated during the invalid time period. The total air volume and the air volume of each zone use the same volumetric flow rate unit, and the filter pressure difference is uniformly converted to Pascals.

[0024] The mapping between zones and injection valves is stored in configuration records. Each configuration record includes a zone identifier, injection valve identifier, control output address, execution order of multiple injection valves within the same zone, single valve pulse duration, and interval between adjacent valve actions. When a zone corresponds to multiple injection valves, the target injection command triggers each injection valve sequentially according to the configuration record, and the time period from the start of action of the first injection valve to the end of action of the last injection valve is taken as one target injection for that zone. The controller records the actual start and end times after receiving valve position feedback or control readback signals; when valve position feedback is not configured, the rising and falling edges of the control output are used as the start and end times.

[0025] Resistance baseline data are established separately for each total airflow operating range. During initial calibration, the controller selects initial non-jet calibration samples based on fan control commands, production equipment status records, jet valve action records, and sensor status records: within the corresponding time period of the sample, the fan control commands and production equipment operating status remain in the same recorded state, there are no jet actions or valve switching, and the airflow sensor and differential pressure sensor have no faults, missing values, or saturation markers. The identification of this sample set does not use ungenerated zonal deviation data, stability thresholds, or dust removal trigger thresholds. The controller divides the allowable total airflow range of the equipment into multiple adjacent intervals. Within each total airflow operating range, the median of the airflow and filtration differential pressure of each zone of the initial non-jet calibration sample is statistically analyzed to form the initial baseline airflow and initial baseline filtration differential pressure. The median of the total airflow in that interval is saved as the representative value of the baseline total airflow.

[0026] The stability threshold is the upper limit of the absolute value of the total airflow change rate. The controller first reads the upper empirical quantile based on the distribution of the absolute change rate of the total airflow in the initial non-jet calibration samples and the allowable frequency of non-jet misjudgments. Then, it takes the larger value between this empirical quantile and the resolution limit obtained by dividing the minimum resolvable change of the airflow sensor by the sampling interval as the stability threshold. After the differential pressure sensitivity coefficient, zonal deviation data, preset outlet airflow threshold, and preset differential pressure threshold are formed, the controller deletes samples from the initial non-jet calibration samples whose total airflow change rate exceeds the stability threshold or whose zonal deviation exceeds the limit. It then recalculates the baseline airflow, baseline filter differential pressure, airflow distribution coefficient, and differential pressure sensitivity coefficient using the retained samples. The same rules are followed to delete newly identified samples exceeding the limits and recalculate until no new samples are added. Since only samples are deleted in each round without re-adding deleted samples, this verification process can be completed within a finite sample set. When the retained samples are empty, the corresponding total airflow operating condition interval is marked as a calibration interval to be supplemented.

[0027] The width of the total air volume operating range is progressively subdivided using debugging data: linear fitting is performed on the reference air volume of each zone within each proposed range. When the maximum fitting residual of any zone exceeds the maximum measurement uncertainty of the corresponding air volume sensor, the range is further divided in half until all fitting residuals do not exceed this uncertainty. Each sub-range must contain different total air volume sample values ​​that make the linear fitting equation solvable. If further subdivision causes any sub-range to fail to meet this condition, subdivision is stopped, and the sub-ranges with residuals still exceeding the limit are marked as supplementary calibration ranges, which are not used for automatic continuous dust removal until supplementary samples are obtained. If the current measured total air volume or the projected total air volume at the time to be executed is located between two representative values ​​of the reference total air volume in adjacent total air volume operating ranges, then linear interpolation is performed on the reference air volume of the corresponding zone within the adjacent total air volume operating range according to the distance from the corresponding total air volume value to the two representative values ​​of the reference total air volume. Among them, the current measured total air volume is used for current operating condition parameter processing, and the projected total air volume at the time to be executed is used for dust removal trigger verification before the next dust removal action is executed. The same stability threshold is used for the non-blowout stability range in the validity judgment of blowout events, so that the selection of benchmark samples and the validity judgment of blowout events have a consistent basis.

[0028] The controller calculates the distribution coefficient of each zone to the total air volume variation based on the baseline air volume of each zone. The calculation formula is as follows: ; In the formula, Representing the The air volume distribution coefficient for each zone is a dimensionless quantity. Representing the The baseline airflow for each zone within the current total airflow operating range, expressed in cubic meters per second. This represents the number of zones. When the sum of the baseline airflow for all zones is positive, the sum of the airflow distribution coefficients for each zone is a unit value; when the sum of the baseline airflow is not positive, the current operating range is marked as invalid and data is collected again. This coefficient is used to project the total airflow change into a change component with the same dimensions as the airflow output of a single zone.

[0029] After the initial baseline, airflow distribution coefficient, differential pressure sensitivity coefficient, and deviation threshold are verified, isolated pulse-jet cleaning is performed on each zone as the target area. During each isolated pulse-jet cleaning, the controller first generates a global fluctuation component of the outlet airflow in the target area before the pulse-jet cleaning based on the average rate of change of total airflow before the start of the cleaning and the airflow distribution coefficient. Then, this component and the baseline value at the start of the cleaning process are subtracted from the outlet airflow in the target area during the cleaning response period to obtain the target airflow recovery trajectory. Multiple valid trajectories within the same zone and the same total airflow operating range are aligned according to the start of the cleaning process, and the median value is taken at each sampling time to generate the airflow recovery curve after the initial cleaning. This curve stores the airflow change in cubic meters per second and can be directly added to the predicted baseline value.

[0030] The preset safety upper limit is determined jointly based on the air outlet branch of the zone, the allowable filtration velocity of the filter bags, and the range of the airflow sensor. The controller obtains the rated allowable airflow of the equipment and the maximum stable airflow verified during the commissioning phase, takes the smaller of the two, and then subtracts the maximum measurement uncertainty of the airflow sensor and the reserved safety margin determined according to the equipment safety procedures to obtain the preset safety upper limit for the corresponding zone. The rated allowable airflow of the equipment comes from the equipment design documents, the maximum stable airflow comes from the commissioning records with no alarms, no filter bag abnormalities, and no sensor saturation, and the reserved safety margin is converted into the zone airflow by the difference between the allowable filtration velocity and the rated filtration velocity in the equipment safety procedures.

[0031] After completing the above data acquisition, mapping, and calibration processes, the controller obtains the total air volume and operating parameters of each zone with a unified time scale, the mapping between the zone and the blow valve, and the resistance reference data that can be read by zone and total air volume operating range.

[0032] S2. Non-jetting global fluctuation elimination, zonal deviation data generation, and target area determination: Non-purge periods are selected between two adjacent purge actions. Their start point is no earlier than the moment when the response amplitude of the previous purge decays to the response significance threshold, and their end point is no later than the start point of the next purge. The controller simultaneously excludes sampling points corresponding to fan start / stop, sensor failure, valve switching, and concurrent purges in other zones. The preset period length is determined from completed continuous valid historical segments: the controller incrementally increases the candidate period length according to the sampling period, calculates the global fluctuation components of the outlet air volume and the global fluctuation components of the filter pressure difference corresponding to adjacent candidate lengths, and selects the shortest candidate length where the difference between the two types of global fluctuation components does not exceed the minimum resolvable change of the corresponding sensor. During online judgment, only the most recent continuous valid period that has been obtained before the current moment and meets the above length is used, without relying on sampling data after the current moment.

[0033] Using the first valid sampling time of the preset time period as the reference time, the sampled value at the reference time is subtracted from the current sampled value to obtain the change in airflow and the change in filter differential pressure for each zone. For any sampling time after the first valid sampling time, the global fluctuation component of the airflow for any zone is calculated according to the following formula: ; In the formula, Representing the The partition in the _th ... The global fluctuation component of the air volume at each sampling moment, in cubic meters per second; Representing the Total air volume at each sampling time; The total air volume at the reference time; It is obtained by normalizing the baseline air volume of each zone within the current total air volume operating range. The total air volume difference multiplied by the zone air volume allocation coefficient still yields the air volume dimension.

[0034] The global fluctuation component of the filter differential pressure is projected onto the non-jet sensitivity between the total airflow change and the zone filter differential pressure change. The controller calculates the differential pressure sensitivity coefficient for each zone using the least squares method within the non-jet calibration sample of the current total airflow operating range. ; In the formula, Representing the The differential pressure sensitivity coefficient of each zone, in Pascals divided by (cubic meters per second). Representing the The partition in the _th ... The filter pressure difference at each sampling time; and These represent the average total air volume and the average filtration pressure difference of the corresponding zones within the calibration sample, respectively. This represents the number of calibration samples. The formula is derived from a univariate linear least squares projection that explains the synchronous change in non-jet pressure difference using changes in total airflow, without directly adding the pressure difference to the airflow. When the sum of squares of the total airflow deviations is zero, the pressure difference sensitivity coefficient is not calculated; the controller continues to collect non-jet samples with changes in total airflow within this operating range until the sum of squares of deviations is positive.

[0035] The global fluctuation component of the filtered pressure difference at each sampling time for any partition is calculated according to the following formula: ; In the formula, Representing the The partition in the _th ... The global fluctuation component of the filter pressure difference at each sampling time is expressed in Pascals. The pressure difference sensitivity coefficient multiplied by the total air volume difference is expressed as a pressure difference dimension, and therefore can be deducted from the change in filter pressure difference in each zone.

[0036] The controller subtracts the global fluctuation components of the same zone, the same sampling time, and the same parameter type from the zone airflow change and zone filter pressure difference change respectively, to obtain the zone airflow deviation and zone filter pressure difference deviation: ; In the formula, and These represent the deviation of the zone's airflow and the deviation of the zone's filter pressure difference, respectively. and These represent the changes in airflow and filtration pressure difference between zones, respectively. Both differences are calculated within the same dimension. The controller combines the two types of deviation sequences and their statistical values ​​for the same zone within a preset time period to form zone deviation data.

[0037] The preset airflow threshold and preset differential pressure threshold are determined by the sample fluctuation distribution in the initial non-jet calibration samples where the total airflow change rate does not exceed the stability threshold. The controller sorts the absolute values ​​of the two types of zoning deviations according to the allowable frequency of misjudgments due to exceeding the zoning deviation limit, reads the upper empirical quantile, and then takes the larger value between this quantile and the corresponding sensor's minimum resolvable change as the threshold. The allowable misjudgment frequency is obtained by dividing the acceptable number of erroneous target area judgments per unit runtime by the number of valid judgments, and remains constant within the same calibration cycle. The two thresholds are used for verification of the aforementioned initial non-jet calibration samples; since the initial sample selection does not use zoning deviation data, there is no loop definition.

[0038] The controller compares the deviations at each valid sampling time point within a preset time period, and saves the minimum zone airflow deviation and the maximum zone filter pressure difference deviation for each zone within that time period. When the zone airflow deviation at any valid sampling time point is lower than a negative preset airflow threshold, or the zone filter pressure difference deviation is higher than a preset pressure difference threshold, the controller adds the corresponding zone to the dust removal set. A negative zone airflow deviation indicates that the zone has experienced an additional decrease in airflow relative to the global change; a positive zone filter pressure difference deviation indicates that the zone has experienced an additional increase in pressure difference relative to the global change. The airflow excess ratio and pressure difference excess ratio of zones that have not exceeded the limits are both set to zero according to the non-negative truncation relationship described later, so that the zone can still enter the subsequent candidate safety prediction and dust removal priority ranking.

[0039] When the set to be cleaned contains multiple zones, the controller determines the cleaning priority weight based on historical effective cleaning events within the same total airflow operating range. A historical effective cleaning event is defined as an event where the zone deviation data before the pulse-jet is complete, there are no concurrent pulse-jet events or sensor saturation during the pulse-jet response period, the target airflow recovery trajectory can be obtained, and the historical residual coupling change from the pulse-jet execution start point to the preset evaluation interval can be calculated based on the corresponding directional airflow transfer parameters. For each historical effective cleaning event, the zone's pre-pulse airflow shortage and zone filter differential pressure excess are divided by the corresponding thresholds to form the airflow excess ratio and differential pressure excess ratio. The recovery value of the target airflow recovery trajectory at the preset evaluation interval is subtracted from the historical residual coupling change from the pulse-jet execution start point to the preset evaluation interval to obtain the non-negative recovery amplitude caused by this cleaning. This is then divided by the preset airflow threshold to form the dimensionless cleaning benefit.

[0040] ; In the formula, , and Representing the number in sequence The proportion of air volume exceeding the limit, the proportion of pressure difference exceeding the limit, and the dimensionless dust removal benefit of each historical effective dust removal event; This represents the minimum deviation in airflow volume within a preset time period prior to the event's purging. This represents the maximum differential pressure deviation of the zone filter within the same preset time period; and This represents the corresponding preset threshold. This represents the target air volume recovery range after deducting historical residual coupling changes at the preset evaluation interval; This represents the preset evaluation interval, in seconds. All ratios are dimensionless.

[0041] The controller outputs non-negative airflow coefficients and differential pressure coefficients, and uses the dimensionless cleaning benefits of historical effective cleaning events as the fitting target. The fitting error for each set of candidate coefficients is calculated according to the following formula: ; In the formula, and Represents the non-negative candidate coefficients to be determined; This represents the number of historical valid dust removal events included in the current solution; This represents the fitting error. The controller enumerates candidate solutions with both coefficients involved, only the air volume coefficient involved, only the pressure difference coefficient involved, and neither coefficient involved. When both coefficients are involved, the normal equation is used for solving; if the normal equation is invertible or a negative coefficient appears in the solution, the candidate is discarded. When a single coefficient is involved, the participation coefficient is calculated by dividing the sum of the products of the exceedance ratio and the dimensionless dust removal benefit by the sum of the squares of the exceedance ratio; the other coefficient is set to zero; if the sum of the squares is zero, the single-coefficient candidate is discarded. When neither coefficient is involved, the zero vector is used. The controller calculates the fitting error of each available candidate solution and selects the candidate solution with the smallest fitting error as the optimal solution. When the errors are the same, the candidate solution with fewer non-zero coefficients is selected.

[0042] The two non-negative coefficients are normalized to obtain the weights, and the two dimensionless over-limit ratios of the current partition are added according to their corresponding weights to obtain the dust removal priority: ; In the formula, and These represent the weight of the air volume and the weight of the pressure difference, respectively, and their sum is 1. and The values ​​are obtained by dividing the maximum air volume shortage in the current preset time period and the maximum filter pressure difference excess in the current zone by the corresponding threshold. Representing the The cleaning priority of each zone is determined. If the sum of two non-negative coefficients is zero, or if there are no historically valid cleaning events in the current total airflow operating range, both weights are set to 0.5. The initial weights are recalculated and replaced after a new historically valid cleaning event is generated. The weight calculation uses all historically valid cleaning events generated since the most recent calibration or filter bag replacement, and is re-executed after each new valid event. When the set to be cleaned is not empty, the controller determines the zone with the highest cleaning priority as the current target area, and the remaining zones are retained in the set to be cleaned. If priorities are the same, the zone with the larger filter pressure difference deviation is selected first; if still the same, the zone with the longer time since the last cleaning is selected. When the set to be cleaned is empty, the current target area is not generated, the current cleaning determination ends, and subsequent operating parameters are collected.

[0043] The partition deviation data obtained by the above processing is used to determine the target area and calculate the dust removal priority of the candidate partition; before the actual blowing of the next target area, the detrended measured air volume projected to the time to be executed and the corresponding reference air volume are used to verify the dust removal trigger.

[0044] S3, Target jetting control, pre-jetting global fluctuation removal, and jetting event response data generation: The controller locates all the corresponding blow valves for the target area based on the mapping between zones and blow valves, and generates a target blow command. The target blow command includes at least the target area identifier, blow valve identifier, control output address, planned start time, single valve pulse duration, and execution order within the same zone. The actuator can be a programmable logic controller output module or a blow valve electromagnetic pilot mechanism. The actuator activates the corresponding blow valve according to the target blow command and returns the actual activation time to the controller.

[0045] The controller determines the start time of the first injection valve's actual activation as the injection execution start point. Within a preset length before the injection execution start point, the total air volume and the air volume of each zone are extracted according to a uniform sampling time, forming a synchronous air volume sampling sequence; simultaneously, the filter pressure difference of each zone is aligned according to the same sampling time, forming a filter pressure difference sequence corresponding to the synchronous air volume sampling sequence. The synchronous air volume sampling sequence mentioned in the claims is used for averaging the air volume changes, and the filter pressure difference sequence is not summed or averaged with the air volume. The controller averages the adjacent sampling differences of the total air volume in the synchronous air volume sampling sequence to obtain the average rate of change of the total air volume before injection: ; In the formula, Represents the average rate of change of total air volume before the start of the injection process, in cubic meters per square second; This represents the sampling interval, in seconds; This represents the number of sampling points in the synchronous airflow sampling sequence. This formula averages the change rates of adjacent samples, representing one implementation of the averaging process described in the claims. If the synchronous airflow sampling sequence cannot form a difference between adjacent samples or the sampling interval is not positive, the blowing event is marked as an invalid event.

[0046] For any elapsed time after the start of the pulse jet execution, the controller generates the global fluctuation component of the air volume before pulse jet and the global fluctuation component of the filter pressure difference before pulse jet for each zone according to the following formula: ; In the formula, This represents the elapsed time from the start of the injection process to the current response sampling point, in seconds. and These are the global fluctuation components before the pulse jet, corresponding to the zone's airflow and zone's filter pressure difference, respectively. The former is in cubic meters per second, and the latter is in Pascals. This vector formula ensures that each operating parameter during the response period only deducts components of the same dimension.

[0047] The controller defines the first preset window from the start of the pulse jet execution to the end of the last pulse jet as the pulse jet response period. For each sampling point within the response period, the measured zone airflow is subtracted from the corresponding global fluctuation component of the airflow before the pulse jet, and the measured zone filter pressure difference is subtracted from the corresponding global fluctuation component of the filter pressure difference before the pulse jet, to obtain the instantaneous response values ​​of the airflow and filter pressure difference, respectively, which are then arranged into a sequence of instantaneous response values ​​in chronological order.

[0048] For each instantaneous response value sequence, an odd-length moving average window is used for filtering. The sampled value at the center of the window is replaced by the arithmetic mean of that sampled value and the equal number of sampled points before and after it; the two ends of the sequence are calculated using the number of sampled points actually available. The controller attempts to calculate from each candidate odd-length window with a sampled number less than the number corresponding to the shortest response delay: first, the maximum absolute amplitude of the unfiltered sequence of the effectively isolated jetting event relative to the baseline value is used as the reference amplitude; then, the candidate window is used for filtering the same event, and the maximum absolute amplitude after filtering is obtained; only candidate windows whose difference between the two amplitudes does not exceed the maximum measurement uncertainty of the air volume sensor are retained, and the window with the smallest mean noise variance before jetting of all calibrated events is selected from the retained candidate windows. The selected window length is written to the configuration record. The filtered air volume response sequence of each zone and the filtered differential pressure response sequence together constitute the jetting event response data.

[0049] If, during the pulse jet response period, a fan starts or stops, another area is pulsed, the total air volume suddenly changes beyond the non-pulse jet stability range, or the sensor becomes saturated or continuously misses sampling, the pulse jet event is marked as an invalid event. Invalid events retain their original records for fault tracing but are not used to update resistance baseline data or directional airflow transfer data.

[0050] S4. Generation of directional airflow transfer data and updating of resistance baseline data: like Figure 2 As shown, the controller traverses the airflow response sequence of each non-target area in the jetting event response data. For each non-target area, the filtered response value corresponding to the starting point of the jetting execution is determined as the baseline value; subsequent sampled values ​​are compared with their adjacent sampled values ​​before and after, local maxima and local minima are identified, and the prominence of each feature point relative to the baseline value is calculated. Only feature points with a prominence greater than the response significance threshold and a continuous sampling number reaching the minimum continuous number are considered as valid peaks or troughs. The response significance threshold is obtained by sorting the absolute values ​​of the airflow noise before the starting point of the jetting execution, reading the upper empirical quantile according to the allowable noise misidentification frequency, and then taking the larger value of the sensor's minimum resolvable change. The minimum continuous number is obtained by dividing the sensor's specified stabilization time by the sampling interval and rounding up, ensuring that there are adjacent sampling points on both sides of the feature point that can be used for local extreme value comparison.

[0051] When multiple valid peaks and troughs exist in a non-target area, the feature point with the largest absolute change relative to the baseline value is selected as the primary feature point; the remaining feature points are saved as quality inspection information. The time difference between the primary feature point and the starting point of the injection is the response delay parameter, and the absolute value of the difference between the primary feature point and the baseline value is the response change magnitude parameter. When the primary feature point is higher than the baseline value, the response change direction parameter is set to positive; when it is lower than the baseline value, it is set to negative. The unit of the response change magnitude parameter is cubic meters per second, and the unit of the response delay parameter is seconds. If there are feature points after the primary feature point with opposite directions and whose magnitude exceeds the response significance threshold, it indicates that the response cannot be represented by a first-order decay process in a single direction. The controller marks this directed partition as not updatable and uses the most recent valid parameter.

[0052] The response tail segment after the principal feature point is fitted using a first-order exponential decay model. This model is derived from approximating the cross-zone airflow disturbance after the short-term jetting as a first-order steady-state process, whose free response satisfies the differential relationship that the rate of change of the response amplitude is negatively proportional to the current response amplitude; integrating this relationship yields the exponential decay form. For the moment after the principal feature point where the response amplitude is still higher than the noise threshold, the response model is: ; In the formula, Representing the After spraying the target area, at a certain time For the The signed air volume transfer caused by a non-target area, in cubic meters per second; The parameter representing the direction of response change takes either a positive or negative unit value from the sign function output. The parameter represents the magnitude of the response change; This represents the response delay parameter; This represents the response decay parameter, measured in seconds. This represents the natural exponential function with the natural constant as its base. The product of the decay parameter and the time difference is dimensionless, and the dimensional relationship of the exponential function holds.

[0053] To obtain the response decay parameter, the principal feature point is defined as the zero point of time difference. The time difference between each effective sampling point and the principal feature point is used as the x-axis. The absolute amplitude of each effective sampling point relative to the baseline value is divided by the response change amplitude parameter to obtain the dimensionless amplitude ratio. The natural logarithm of the dimensionless amplitude ratio is taken, and a least-squares fit passing through the origin is performed according to the following formula: ; In the formula, Representing the The time difference between each valid sampling point and the main feature point is expressed in seconds. Represents the absolute response magnitude of the sampling point relative to the baseline value, in cubic meters per second; This represents the number of valid sampling points used for fitting. The obtained value is a dimensionless ratio. The unit is per second. When the number of effective sampling points does not reach the minimum fitted sample number obtained by dividing the sensor's specified settling time by the sampling interval and rounding up, the sum of squares of the time differences is zero, or the fitted attenuation parameter is not greater than zero, the directed partition pair is marked as not updatable for this time, and its most recent effective parameter is used. After the attenuation parameter meets the above conditions, the controller also uses the fitting model to calculate the fitting amplitude of each effective sampling point, squares the difference between the measured absolute amplitude and the fitted amplitude in turn, and takes the square root of the mean of the square differences as the fitting residual. The effective fitting threshold is the larger value between the maximum measurement uncertainty of the air volume sensor and the empirical quantile on the upper side of the absolute amplitude of the noise before the start of the jet execution; when the fitting residual does not exceed this threshold, the curve fitting is considered effective; otherwise, the directed partition pair is marked as not updatable for this time.

[0054] The controller encapsulates the target area identifier, non-target area identifier, response change direction parameter, response change amplitude parameter, response delay parameter, response attenuation parameter, total airflow operating range, event time, and validity status into a single directional airflow transfer record. For the same directional zone pair, among all valid records formed since the most recent calibration or filter bag replacement, the direction parameter uses the majority direction of the valid records, while the amplitude, delay, and attenuation parameters use the median of the valid records. If the number of positive and negative records is the same, the most recently confirmed direction is used. Therefore, the starting point, member range, and statistical method of the aggregated record can all be recalculated from the event log.

[0055] Simultaneously, the filtered airflow response sequence of the target area is extracted from the jetting event response data, and the baseline value of the jetting execution start point is subtracted from each sampling point to generate the target airflow recovery trajectory. The controller reads the post-jet airflow recovery curves of the same zone and the same total airflow operating range, calculates the difference point by point at sampling points with the same elapsed time, generates a recovery deviation sequence, and updates the curves according to the following formula: ; In the formula, and These represent the numbers before and after the update, respectively. The air volume recovery value after a certain period of time; This represents the observed recovery value of the target air volume recovery trajectory at the corresponding time. The update coefficient representing the recovery curve is a dimensionless quantity that satisfies the convex combination constraint. This formula represents the convex combination of the old curve and the current observation, derived from the online exponential weighted update relationship, and does not change the dimensions of the wind volume.

[0056] The recovery curve update coefficients are solved from historical valid events according to the event quality level. Events with complete injection response periods, total air volume changes within the calibration range, and effective curve fitting are classified into the high-quality event set; events with interpolation sampling but complete main feature points and attenuation segments are classified into the medium-quality event set. For each event set, according to the order of event occurrence, the curve obtained after updating a historical event is used as the prediction curve for the next historical event. The squared deviation between the observed trajectory and the prediction curve of the next historical event at the corresponding sampling time is accumulated. Among the candidate coefficients that satisfy the convex combination constraints, the one with the smallest accumulated squared deviation is selected, and if the errors are the same, the smaller candidate coefficient is selected. The candidate coefficients traverse the entire feasible interval of the convex combination according to the smallest parameter resolution step size that the controller can save. Invalid events are not updated. The obtained coefficients, event set identifiers, accumulated errors, and the data version used for solving are written together into the configuration record, and the corrected post-injection air volume recovery curve is written into the resistance baseline data.

[0057] S5. Zonal airflow prediction, candidate zonal screening, and determination of the next target area: like Figure 3 As shown, the execution time is no earlier than the moment when the airflow recovery curve of the previous target area enters the stable interval. The stable interval continuously covers the sensor's specified stabilization time, and the absolute value of the curve's rate of change does not exceed the recovery stabilization threshold; this threshold is the larger of the upper empirical quantile of the absolute value of the airflow noise change rate before injection and the airflow change rate resolution limit. The residual coupling amount only accumulates the directional airflow transfer from the historical target area to the non-target area, and no longer superimposes the historical target area's own recovery curve, because its stabilized state is already included in the measured airflow at the execution time. The predicted evaluation time is the execution time plus a preset evaluation interval; this interval is taken as the upper empirical quantile of the response delay of each directed partition according to the allowable response omission frequency.

[0058] The controller retrieves the most recent historical dust removal actions from the time to be executed backwards. To determine the tracing range, it reads the maximum response change amplitude, maximum response delay, and minimum positive response decay parameters within the current calibration data version, constructing a theoretical residual upper bound that monotonically does not increase with the elapsed time from the start of the pulse jet: if the elapsed time does not exceed the maximum response delay, the maximum response change amplitude is used; if it does, it decays exponentially according to the minimum positive response decay parameter. Starting from the most recent action, the controller compares this upper bound with the residual neglect threshold in reverse order. Tracing stops when the upper bound first does not exceed the threshold, and subsequent actions are included in the calculation. The residual neglect threshold is the larger value between the minimum resolvable change of the airflow sensor and the empirical quantile above the stable tail absolute response amplitude of the effective pulse jet event.

[0059] For any historical dust removal action included in the calculation and any affected zone, the controller reads the response change direction parameter, response change amplitude parameter, response delay parameter, and response decay parameter for the corresponding directed zone pair. The time offset between the time to be calculated and the end time of the action is added to the duration of the pulse jet to obtain the total elapsed time from the start of the pulse jet execution to the time to be calculated; then, the response delay parameter is subtracted to obtain the decay time after the response arrives. When this time is less than zero, the residual air volume influence is zero; when it is not less than zero, the increase / decrease direction is represented by the response change direction parameter, the initial amplitude at arrival is used by the response change amplitude parameter, and the calculation is performed according to the response decay parameter. ; In the formula, Representing the This historical cleansing action is timely. For the The residual air volume impact of each zone, in cubic meters per second; Representing the The target area for the previous historical dust removal operation; Representative moment The time offset from the end time of the historical dust removal operation, in seconds; This represents the duration of the historical dust removal operation from the start to the end of the blowing process, in seconds; , , and Representing from the first The target area to the first The parameters for the direction of response change, the magnitude of response change, the response delay, and the response attenuation of each affected partition are defined in the formula. This represents a natural exponential function with the natural constant as its base. The dimensionless exponent is obtained by multiplying the response decay parameter by the decay time after the response arrives; the negative sign before the exponent indicates that the residual amplitude decreases as the decay time increases. The response change direction parameter retains the direction of airflow increase or decrease, therefore the residual airflow effect can be positive or negative.

[0060] The current residual coupling amount for a given zone is obtained by summing the impact of each historical dust removal action on the residual air volume of the same zone at the time of execution; the evaluated residual coupling amount for a given zone is obtained by summing the impact of each historical dust removal action on the residual air volume of the same zone at the time of prediction and evaluation. ; In the formula, and Representing the first The current residual coupling and the evaluation residual coupling of each partition; and These represent the time to be executed and the time to predict and evaluate, respectively. This represents the number of historical dust removal actions included in the calculation. Positive and negative directions are preserved during summation, so that the opposing effects of different historical actions on the same partition can cancel each other out.

[0061] The controller subtracts the current residual coupling of the corresponding zone from the measured zone airflow at the time to be executed, and generates the prediction baseline value: ; In the formula, Representing the The forecast baseline value for each partition; This represents the measured airflow rate of the zone at the time of execution. The prediction baseline, measured airflow rate, and current residual coupling are all expressed in cubic meters per second.

[0062] Before executing candidate predictions, the controller performs a least-squares linear fit between the total air volume of the most recent consecutive effective non-jetting period before the execution time and the sampling time, and extrapolates the resulting slope to the prediction evaluation time to obtain the projected total air volume. Prediction continues as long as the total air volume at the execution time and the prediction evaluation time are both within the same operating range and the rate of change does not exceed the stability threshold. When the projected total air volume is between adjacent representative values ​​of the baseline total air volume, linear interpolation is performed on the baseline data based on its distance to the two representative values. If the rate of change exceeds the limit or the projected total air volume leaves the available calibration range, the controller waits for the total air volume to stabilize again, sets the new current time as the execution time, and re-determines the evaluation time, operating range, and prediction baseline values ​​for each zone.

[0063] The controller sequentially selects each zone outside the current target area as a candidate zone and reads the post-purge airflow recovery curve of the corresponding operating condition interval and the directional airflow transfer data from the candidate zone to each non-candidate zone. For any non-candidate zone, if the preset evaluation interval is less than the response delay parameter, the predicted transfer amount is zero; otherwise, the predicted transfer amount is the product of the response change direction parameter, the response change amplitude parameter, and the natural exponential decay term, where the exponent is the negative value of the product of the response decay parameter and the time difference obtained by subtracting the response delay parameter from the preset evaluation interval. In automatic continuous cleaning mode, all zones have been calibrated, and the candidate list is empty, generated only by the predicted airflow safety screening. If the calibration data of any candidate zone fails, automatic continuous cleaning is paused, its non-purge calibration sample is re-collected, and isolated purge calibration is performed without entering the delay recalculation branch.

[0064] For each candidate zone, the predicted baseline value, the evaluated residual coupling amount, and the recovery value of the post-injection airflow recovery curve at the preset evaluation interval are summed; for each non-candidate zone, the predicted baseline value, the evaluated residual coupling amount, and the corresponding predicted transfer amount are summed. ; ; In the formula, Representative of the first The predicted air volume of a candidate zone when performing dust removal; Representing the first candidate in the same candidate situation Predicted air volume for each non-candidate zone; This represents the recovery value of the air volume recovery curve after the injection at the preset evaluation interval; Representing the After the first candidate zone is sprayed, the second... The predicted transfer volume for each non-candidate zone. All summation terms are air volume, with consistent dimensions.

[0065] The exclusion of zones with predicted airflow exceeding limits, as referred to in this manual, means that if any zone in the set of predicted airflow values ​​for a candidate zone exceeds its preset safety limit, that candidate zone will be excluded from the action; it does not simply delete affected zones predicted to exceed limits from the prediction results. Candidate zones with no predicted airflow exceeding limits are added to the candidate list.

[0066] The controller divides the maximum airflow shortage and maximum filtration differential pressure excess of each candidate zone within the current preset time period by their respective thresholds. Then, it weights these two values ​​according to two categories of weights derived from historical effective cleaning events to determine the cleaning priority, selecting the zone with the highest priority as the next target area. Candidate zones where neither type of deviation exceeds the limit have a zero excess ratio for either type, but they still participate in safety prediction. They are only selected when no higher-priority candidates exist. Before the pulse jet, the difference between the mean of the detrended measured airflow sequence projected to the execution time and the baseline airflow corresponding to the total projected airflow at the execution time is used to verify whether cleaning should continue. Therefore, safety screening takes precedence over cleaning urgency ranking, meaning that any candidate zone with a predicted airflow exceeding the limit cannot become the next target area.

[0067] S6. Delayed recalculation and dust removal blocking handling when the candidate list is empty: The controller checks if the candidate list is empty. If the candidate list is empty, it means that performing pulse-jet cleaning on any candidate zone at the current moment may cause the predicted airflow of at least one zone to exceed the preset safety limit. The controller does not immediately change the preset safety limit, nor does it force the selection of the candidate zone with the highest cleaning priority. Instead, it initiates a delayed recalculation mechanism.

[0068] The delay recalculation interval is set as an integer multiple of the sampling period and is not less than the shortest response delay in the current effective directional airflow transfer records. When a historical empty candidate list event exists, the calibration phase progressively increases the delay starting from one sampling period, and replays the historical candidate selection process at each delay. For each historical empty candidate list event, the delay required for the first reappearance of a non-empty candidate list is recorded, and the median of all required delays is taken and adjusted upwards to an integer multiple of the sampling period as the initial delay recalculation interval. When no historical empty candidate list event exists, the larger of the sensor's specified settling time and the shortest response delay in the current effective directional airflow transfer records is taken and adjusted upwards to an integer multiple of the sampling period as the initial delay recalculation interval. After a historical empty candidate list event is formed, the initial value is replaced by the historical playback results. Therefore, a definite delay recalculation interval can be obtained even when the candidate list is empty for the first time.

[0069] After the delay ends, the controller resets the current time as the execution time, adds the original preset evaluation interval to the new execution time to obtain the new predicted evaluation time, recalculates the time offset of each historical dust removal action relative to the two times, and then recalculates the current residual coupling, the evaluated residual coupling, the predicted baseline value, and the predicted air volume value for each candidate zone. If the regenerated candidate list is not empty, the delay recalculation ends, and the next target area is determined according to the dust removal priority.

[0070] The preset maximum retry threshold is determined by dividing the maximum blockage time of a cleaning batch allowed by the equipment safety regulations by the delay recalculation interval and rounding down. If the candidate list remains empty after each recalculation and the number of retries has not exceeded the threshold, the delay continues. If the threshold is exceeded, the controller will automatically lock the pulse-jet output, stop sending new target pulse-jet commands, and generate a cleaning blockage warning signal including the blockage start time, the set of areas to be cleaned, the over-limit prediction results for each candidate zone, and the number of retries. The locked state is released by manual confirmation, a working condition recovery signal, or a recalibration completion signal.

[0071] S7. Next target area dust removal trigger verification, subsequent pulse-jet cleaning and prediction deviation correction: Before the actual pulse-jet cleaning of the next target area, the controller determines the action to be executed as the next dust removal action, and defines the second preset window that is consecutive and immediately adjacent to the execution time as the reference period before the next dust removal. It then reads the total air volume and the zone air volume corresponding to the next target area within this reference period. The end of the reference period before the next dust removal is the last valid sampling time before the execution time. When the execution time is reset due to delay recalculation or changes in operating conditions, the controller re-determines the reference period before the next dust removal and recalculates the linear change trend of the total air volume and the dust removal trigger deviation value. The discernible timescale of filter bag dust accumulation is obtained from long-term non-pulse-jet historical data: for each valid historical segment, the change in zone outlet air volume and the change in zone filtration pressure difference are obtained with the first valid sampling time as a reference; the total air volume change is multiplied by the air volume distribution coefficient and the pressure difference sensitivity coefficient to obtain two types of global fluctuation components of the same dimension, which are then subtracted from the corresponding changes; the elapsed time when any zone deviation first reaches the corresponding threshold is recorded, and the minimum value among all records is taken. If no identifiable change record is available, the shortest dust accumulation state change cycle in the equipment maintenance record is used as the alternative upper limit. The second preset window should be shorter than this time scale and include non-coincident sampling times that allow the least squares equation for a linear trend in total air volume to be solvable; the controller selects the longest duration from the candidate windows that meet the conditions.

[0072] The controller uses least-squares linear fitting to measure the total air volume within the reference time period and calculates the slope of the linear trend of the total air volume: ; In the formula, The slope representing the linear trend of total air volume, expressed in cubic meters per square second; Representing the Each sampling time; This represents the average value of the sampling times within the reference period; This represents the average total air volume over the reference period. The formula is derived from the ordinary least squares relation that minimizes the squared error between the sampled total air volume and the linear trend.

[0073] The controller determines the execution time of the next dust removal action as the unified projection reference time. Based on the linear change trend of the total air volume during the reference period before the next dust removal, the fitted value of the total air volume at each sampling time and the projected value at the execution time are calculated respectively. The difference between the two is then projected according to the air volume distribution coefficient of the next target area to generate the global fluctuation component before the next dust removal. ; In the formula, The next target area is in The global fluctuation component before the next dust removal at each sampling time relative to the time to be executed; The air volume distribution coefficient represents the air volume distribution coefficient for the next target area; The slope representing the linear trend of total air volume change; Representing the Each sampling time; This represents the time to be executed before the next dust removal action. The air volume distribution coefficient is a dimensionless quantity. The slope of the linear change trend of the total air volume multiplied by the time difference gives the air volume dimension. Therefore, the global fluctuation component before the next dust removal can be subtracted from the zone air volume.

[0074] The controller subtracts the corresponding global fluctuation component before the next dust removal from the zone air volume of the next target area at each sampling time to generate a detrended measured air volume sequence: ; In the formula, The next target area is in Detrended measured air volume after projecting each sampling time to the time to be executed; The measured zone airflow represents the next target area. This processing will uniformly correct the zone airflow differences caused by the linear change of total airflow at each sampling time within the reference period to the total airflow level at the time to be executed, while retaining the local airflow deviation of the next target area itself.

[0075] The controller calculates the projected total air volume at the time of execution based on the linear trend of the total air volume change. ; The controller reads the reference air volume corresponding to the projected total air volume from the resistance reference data. When the projected total air volume is between adjacent representative values ​​of the reference total air volume, the reference air volume of the corresponding zone is linearly interpolated according to the distance between the projected total air volume and the two representative values ​​of the reference total air volume to obtain the reference air volume corresponding to the time to be executed. The controller calculates the arithmetic mean of the detrended measured air volume sequence and subtracts the reference air volume corresponding to the time to be executed to generate the dust removal trigger deviation value. ; In the formula, This represents the dust removal trigger deviation value for the next target area; This represents the number of valid sampling points within the reference time period before the next dust removal; This represents the reference air volume corresponding to the projected total air volume at the time of execution in the resistance reference data. A negative dust removal trigger deviation value indicates that the next target area is still lower than the corresponding reference air volume after excluding the influence of linear changes in total air volume.

[0076] The preset trigger threshold is determined by non-trigger reference samples independent of the dust removal trigger determination result. Non-trigger reference samples are selected from the initial non-jet calibration samples and subsequent valid non-jet periods. The selection criteria are: the total airflow change rate does not exceed a stable threshold, the zone outlet airflow deviation and zone filter pressure difference deviation do not exceed their respective preset thresholds, there are no pending jetting commands, and the sensor status is valid. For each non-trigger reference sample, the controller projects the zone airflow within the corresponding reference period to the corresponding determination time, reads the baseline airflow corresponding to the projected total airflow at that determination time, and calculates the difference between the mean of the detrended measured airflow sequence and the baseline airflow to obtain the dust removal trigger deviation of the non-trigger reference sample. The controller calculates the absolute value of each dust removal trigger deviation, reads the upper empirical quantile of the absolute value distribution according to the allowable frequency of dust removal trigger misjudgments, and takes the larger value between the empirical quantile and the minimum resolvable change of the airflow sensor as the preset trigger threshold. Therefore, the non-trigger reference sample is not identified based on the yet-to-be-determined preset trigger threshold.

[0077] When the dust removal trigger deviation value is not greater than the negative preset trigger threshold, the controller sets the next target area as the new target area, finds the corresponding jet valve according to the mapping between the zone and the jet valve, generates and sends the target jet command. After the jet is completed, the global fluctuation component before the jet is generated using the synchronous air volume sampling sequence before the jet execution start point. The same dimension component is subtracted from the operating parameters of each zone during the jet response period, and a new jet event response data is formed by moving average filtering. When the dust removal trigger deviation value does not meet the above conditions, the current round of continuous dust removal ends. Local resistance anomaly identification uses both the zone outlet air volume deviation and the zone filter pressure difference deviation. The review before the actual jet is based on the still existing zone air volume shortage as the final condition for continuing the jet. When only the filter pressure difference deviation exceeds the limit, the pressure difference anomaly record is retained for subsequent judgment, and the jet does not continue in the current round.

[0078] After a new target area is purged, the controller reads the filtered airflow response values ​​of each zone at the preset evaluation interval, as well as the predicted baseline values ​​and predicted airflow values ​​of each zone saved when the target area was used as a candidate zone for prediction. The actual response value is obtained by subtracting the corresponding predicted baseline value from the filtered airflow response value, and the predicted response value is obtained by subtracting the same predicted baseline value from the predicted airflow value; the same zero point ensures that historical residual changes are counted according to the same caliber. If the event is invalidated due to fan start / stop, concurrent purging, sudden change in total airflow, sensor saturation, or continuous missing sampling during the purging response period, no data correction is performed.

[0079] The controller subtracts the predicted response value from the actual response value to obtain the prediction deviation for the target area and the prediction deviation for each non-target area. For a new target area, the recovery value of the original recovery curve at the preset evaluation interval is added to the product of the recovery curve update coefficient and the prediction deviation for the target area to obtain the corrected recovery value. Linear interpolation of adjacent sampling points is used to maintain curve continuity. The target airflow recovery trajectory generated using the baseline value at the start of the injection execution is used to update the airflow recovery curve point-by-point after injection. The prediction reference value serves as the common zero point for both the actual and predicted responses, used to correct the prediction error at the prediction evaluation time. When the preset evaluation interval is not less than the corresponding response delay parameter, the prediction deviation for non-target areas corrects the response change magnitude parameter of the corresponding directed partition pair according to the following formula: ; In the formula, and These represent the magnitude parameters of the response change before and after the correction, respectively. Representing the Non-target area prediction bias; The directional air volume transfer data update coefficient is a dimensionless quantity that satisfies the convex combination constraint. This represents the preset evaluation interval. The exponent term is used to convert the decayed deviation at the evaluation time to the magnitude scale of the principal feature point; the update result is restricted to non-negative values, and the positive and negative directions of the response are still saved by the response change direction parameter.

[0080] When the preset evaluation interval is less than the corresponding response delay parameter, and the evaluation time has not yet reached the effective response interval of the directed partition pair, the controller does not use the above amplitude correction formula. Instead, it re-extracts the response change amplitude parameter after the complete response sequence reaches the main feature point. The update coefficient of the directional air volume transfer data and the update coefficient of the recovery curve are obtained using the same historical event classification and cumulative prediction error minimization method. If the absolute value of the prediction deviation of the non-target area does not exceed the response significance threshold, the corresponding parameter is not corrected; if it exceeds the threshold, amplitude correction is performed, and the observed response delay parameter and observed response attenuation parameter are re-extracted from the current complete response sequence and then weighted with the old parameters according to the same update coefficient. Only when the main feature points opposite to the original direction are in the majority of the effective events formed since the most recent calibration or filter bag replacement, and their median amplitude exceeds the response significance threshold, is the response change direction parameter switched to the opposite direction to avoid direction reversal caused by single noise.

[0081] The exponential decay formula, least squares trend formula, residual superposition formula, and online weighted update formula in this embodiment are all directly derived from the data definitions or first-order response assumptions described in this specification. The core inventive point lies in using the direction, amplitude, delay, and attenuation of directed partition pairs to calculate the historical residual effects at two time points, and jointly predicting the air volume of each partition with the target recovery and cross-partition transfer of candidate jets, rather than taking a general mathematical formula as an independent inventive point. Therefore, all calculations can be reproduced without relying on unpublished training models or external literature.

[0082] The controller saves the following data for each processing cycle: original sampling data, global fluctuation components, zone deviation data, pulse-jet event response data, directional airflow transfer data, resistance baseline data version, candidate list, predicted airflow value for each candidate scenario, pending execution time of the next cleaning action, projected total airflow at the pending execution time, corresponding baseline airflow, cleaning trigger deviation value, trigger judgment result, and model correction record. All saved data includes zone identifiers, event identifiers, sampling time, and total airflow operating range, allowing for verification of the complete processing path from input to cleaning control output.

[0083] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. A method for controlling the airflow in zones for cleaning jet valves, characterized in that, include: Obtain the total air volume and operating parameters of each zone, establish the mapping between zones and jet valves, and generate resistance baseline data; Global fluctuations during non-flushing periods are removed from the operating parameters of each zone to generate zone deviation data, and the target area is determined based on the zone deviation data. According to the mapping, the corresponding jet valve of the target area is controlled to perform target jetting, eliminate global fluctuations before jetting, and generate jetting event response data; Based on the jetting event response data, directional airflow transfer data from the target area to the non-target area is generated; The resistance baseline data is updated based on the target area and the jetting event response data. Combining the directional airflow transfer data and the resistance baseline data, the airflow of each zone is predicted. Zones with predicted airflow exceeding the limit are excluded, and the remaining zones are used as candidate zones. The next target area is then determined from the candidate zones. Determine whether the next target area meets the dust removal trigger condition; If the conditions are met, the next target area is set as the new target area, and the process returns to the step of executing target blowing. If the requirements are not met, the dust removal process will end. The start time of the target spraying is determined as the spraying execution start point; Extract the synchronous air volume sampling sequence of a preset length before the starting point of the jetting execution, and perform mean processing on the synchronous air volume sampling sequence to generate a global fluctuation component before jetting. The first preset window from the start of the blowing execution to the end of the blowing pulse is defined as the blowing response period; Subtract the global fluctuation component before injection from the operating parameters of each sampling point during the injection response period to obtain the instantaneous response value sequence; The instantaneous response value sequence is filtered by a moving average to generate the jetting event response data; Traverse the non-target region data subset in the jetting event response data to identify peak and trough feature points; The response value corresponding to the starting point of the jetting execution is determined as the baseline value; Calculate the time difference between the peak and trough feature points and the starting point of the jetting execution to generate response delay parameters; Calculate the difference between the amplitude of the peak and trough feature points and the baseline value to generate a response change amplitude parameter; Based on the amplitude change direction of the peak and trough feature points, a response change direction parameter is generated; The response amplitude sequence after the peak and trough feature points is subjected to exponential fitting, and the attenuation coefficient of the fitted curve is extracted to generate the response attenuation parameter. The directional airflow transfer data is generated by encapsulating the response change direction parameter, the response change amplitude parameter, the response delay parameter, and the response attenuation parameter according to the non-target area index.

2. The method for zoned airflow control for jet valve cleaning according to claim 1, characterized in that, The step of determining the target region based on the partition deviation data includes: Obtain the changes in zone airflow and zone filtration differential pressure during the non-jet blowing phase within a preset time period; Calculate the global fluctuation components of the air volume and the global fluctuation components of the filter pressure difference at each sampling time. The air volume variation of the partition is subtracted from the global fluctuation component of the air volume to obtain the partition air volume deviation. The filter pressure difference variation of the partition is subtracted from the global fluctuation component of the filter pressure difference to obtain the partition filter pressure difference deviation. The partition air volume deviation and the partition filter pressure difference deviation are combined to form the partition deviation data. The deviation of the air volume of the partition is compared with the preset air volume threshold, and the deviation of the filter pressure difference of the partition is compared with the preset pressure difference threshold. If either the deviation of the air volume of the zone or the deviation of the filter pressure difference of the zone exceeds the limit, the corresponding zone will be determined as the target area.

3. The method for zoned airflow control for jet valve cleaning according to claim 1, characterized in that, The step of controlling the jet valve corresponding to the target area to perform target jetting according to the mapping includes: The corresponding jet valve for the target area is located based on the mapping. Based on the target area and its corresponding blow valve, a target blow command is generated to control the blow valve to perform a blow action. The target blowing command is sent to the actuator to control the blowing valve to perform target blowing.

4. The method for zoned airflow control for jet valve cleaning according to claim 1, characterized in that, Determining the next target region from the candidate partitions includes: Determine the time to be executed for the next dust removal action, and determine the predicted evaluation time based on the time to be executed and the preset evaluation interval; Retrieve directional airflow transfer data corresponding to the most recent historical dust removal actions before the time to be executed; Calculate the time offset between the time to be executed and the predicted evaluation time and the end time of each historical dust removal action; Based on the response change amplitude parameter, response delay parameter, response attenuation parameter, and corresponding time offset of each historical dust removal action, calculate the current residual coupling amount of each partition at the time to be executed and the evaluation residual coupling amount at the time of prediction and evaluation. Subtract the accumulated value of the current residual coupling from the operating parameters of each partition corresponding to the time to be executed to generate a prediction baseline value; Each non-target area is sequentially selected as a candidate partition, and the post-jetting air volume recovery curve corresponding to each candidate partition is read from the resistance reference data. Retrieve directional air volume transfer data generated in historical dust removal operations when each candidate zone is the target area, and calculate the predicted transfer amount to each non-candidate zone after dust removal is performed in each candidate zone at the predicted evaluation time based on the corresponding response change direction parameters, response change amplitude parameters, response delay parameters and response attenuation parameters. The evaluation residual coupling amount is superimposed on the prediction benchmark value, the post-jetting air volume recovery curve is superimposed on the candidate partition, and the corresponding prediction transfer amount is superimposed on each non-candidate partition to obtain the prediction air volume value of each candidate partition. Exclude candidate zones whose predicted air volume exceeds the preset safety limit, and add the remaining candidate zones to the candidate list; The dust removal priority of each candidate partition in the candidate list is determined based on the partition deviation data, and the candidate partition with the highest dust removal priority is determined as the next target area.

5. The zoned airflow control method for cleaning jet valves according to claim 4, characterized in that, The method further includes: Determine whether the candidate list is empty; When the candidate list is empty, a delay recalculation mechanism is activated, the current time after the delay is reset as the time to be executed, and the predicted evaluation time is re-determined according to the preset evaluation interval. Based on the reset execution time and prediction evaluation time, the current residual coupling amount and the evaluation residual coupling amount are recalculated, and the predicted wind volume value of each candidate partition is recalculated to regenerate the candidate list. When the regenerated candidate list is not empty, the delayed recalculation mechanism ends, and the next target area is determined from the regenerated candidate list according to the dust removal priority; If the regenerated candidate list is still empty and the number of consecutive delayed recalculations has not exceeded the preset maximum retry threshold, the delayed recalculation mechanism will continue to be executed. When the number of consecutive delay recalculations exceeds the preset maximum retry threshold and the candidate list is still empty, the dust removal output is locked and a dust removal blockage warning signal is generated.

6. The zoned airflow control method for cleaning jet valves according to claim 4, characterized in that, The step of updating the resistance baseline data based on the target area and the jetting event response data includes: Extract the target airflow recovery trajectory corresponding to the target area from the jetting event response data; Read the post-jet airflow recovery curve corresponding to the target area in the resistance reference data; The target air volume recovery trajectory is compared with the recovery value at the corresponding sampling time in the post-jet air volume recovery curve to generate a recovery deviation sequence; The post-jet airflow recovery curve for the target area is corrected based on the recovery deviation sequence. The corrected post-jet airflow recovery curve is written into the resistance baseline data to generate updated resistance baseline data.

7. The zoned airflow control method for cleaning jet valves according to claim 6, characterized in that, The determination of whether the next target area meets the dust removal trigger condition includes: The cleaning action to be performed on the next target area is determined as the next cleaning action; The second preset window before the next dust removal action is executed is determined as the reference time period before the next dust removal; Obtain the total air volume during the reference period before the next dust removal and the zone air volume corresponding to the next target area; Calculate the linear change trend of the total air volume during the reference period before the next dust removal, and project the linear change trend onto each sampling point during the reference period before the next dust removal to generate a global fluctuation component before the next dust removal. Subtract the global fluctuation component before the next dust removal from the partition air volume to generate a detrended measured air volume sequence. Calculate the difference between the mean of the measured air volume sequence and the reference air volume corresponding to the next target area in the resistance reference data to generate the dust removal trigger deviation value; The dust removal trigger deviation value is compared with the preset trigger threshold to generate a trigger determination result; Based on the trigger determination result, determine whether the next target area meets the dust removal trigger condition.

8. The method for zoned airflow control for cleaning jet valves according to claim 7, characterized in that, The method further includes: When the next target area meets the dust removal triggering condition, the next target area is set as a new target area, the blow valve corresponding to the new target area is controlled to perform target blow, and new blow event response data is generated. Extract the actual response values ​​of each partition from the preset evaluation interval at the starting point of the jetting event from the new jetting event response data; Retrieve the predicted air volume values ​​of each partition and the predicted baseline value when the new target area is used as the candidate partition, and generate the predicted response value of each partition based on the difference between the predicted air volume value of each partition and the predicted baseline value. Calculate the target region prediction deviation between the actual response value and the predicted response value corresponding to the new target region, and the non-target region prediction deviation between the actual response value and the predicted response value corresponding to each non-target region. The post-jet airflow recovery curve corresponding to the new target area in the updated resistance baseline data is corrected based on the target area prediction deviation. The directional airflow transfer data from the new target area to the corresponding non-target area is corrected based on the prediction deviation of each non-target area.

Citation Information

Patent Citations

  • Denitration ammonia spraying control method and system based on accurate measurement of ammonia escape

    CN117065570A

  • Cloth bag dust cleaning intelligent control system and method based on model prediction

    CN121731886A