A method for optimizing pneumatic rapping ash efficiency under high-ash condition

Abstract

The application discloses a kind of high ash condition under pneumatic rapping ash efficiency optimization method, it is related to flue gas electrostatic precipitation technical field, including, to high ash condition rapping data set carries out time window frequency feature extraction, obtains rapping multi-modal feature set, quantifies rapping performance state by rapping state evaluation algorithm, generates rapping comprehensive state set;Based on rapping comprehensive state set, construct double-constraint rapping optimization model, by optimizing configuration to rapping intensity parameter, generate regional rapping optimization strategy;Execute regional rapping optimization strategy, combine high ash condition rapping data set by acoustic vibration feature to identify the marginal ash cleaning effect of pneumatic rapping, obtain rapping execution result;According to rapping execution result, double-constraint rapping optimization model is adaptively adjusted, and combined with rapping comprehensive state set carries out ash cleaning performance trend analysis, obtains ash cleaning efficiency optimization report.The application is by constructing rapping comprehensive state set, improves ash cleaning efficiency and reduces unit ash cleaning energy consumption.

Description

Technical Field

[0001] This invention relates to the field of flue gas electrostatic dust removal technology, and in particular to a method for optimizing the efficiency of pneumatic vibration dust removal under high ash conditions. Background Technology

[0002] In the flue gas purification process of waste incinerators and their associated waste heat boilers, electrostatic precipitators are widely used to electrostatically separate suspended particles in the flue gas. Ash gradually accumulates on the collecting plates and discharge wires within the electrostatic precipitator, requiring periodic cleaning via pneumatic rapping devices to maintain electrostatic precipitator efficiency and pressure drop within a reasonable range. Engineering practices have evolved from early reliance on manual hammering and simple mechanical scraping to the use of electrostatic precipitator-compatible collecting plates, discharge wires, and pneumatic rapping devices. Under high ash conditions, this approach breaks down the adhesion between the ash layer and the substrate, allowing the ash to be discharged by gravity and the flue gas, thereby improving heat exchange on the heating surfaces and flue gas flow conditions.

[0003] Under high-ash conditions, existing electrostatic precipitators for flue gas still have two limitations: First, current control strategies are mostly based on simple interlocking of fixed rapping programs or a small number of process quantities, lacking the ability to synchronously quantify the high ash level, structural fatigue margin, and rapping energy consumption of each cleaning zone on a single rapping time scale. This results in some areas being under- or over-vibrated for extended periods, and excessively high unit cleaning energy consumption. Second, existing solutions typically consider the optimization of cleaning effect separately from the constraints of fatigue life of heated surfaces or shells and compressed air energy consumption. They have not yet formed a system with high ash index and structural fatigue margin as dual constraints, making it difficult to make timely and refined adjustments to the regional rapping intensity and rapping strategy based on the rapping execution effect. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for optimizing pneumatic rapping cleaning efficiency under high ash conditions to solve the problems of insufficient perception accuracy and lack of integrated optimization adjustment in high ash conditions and edge cleaning efficiency.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a method for optimizing the cleaning efficiency of pneumatic rapping under high-ash conditions. The method includes: collecting pneumatic rapping operation data and structural acoustic and vibration signals, and preprocessing them to obtain a high-ash-condition rapping dataset; extracting time-frequency features from the high-ash-condition rapping dataset using time windows to obtain a multi-modal feature set of rapping; quantifying the rapping performance state using a rapping state evaluation algorithm to generate a comprehensive rapping state set; constructing a dual-constraint rapping optimization model based on the comprehensive rapping state set; generating a regional rapping optimization strategy by optimizing the rapping intensity parameters; executing the regional rapping optimization strategy; identifying the marginal cleaning effect of pneumatic rapping using acoustic and vibration features in conjunction with the high-ash-condition rapping dataset to obtain rapping execution results; adaptively adjusting the dual-constraint rapping optimization model based on the rapping execution results; and performing cleaning performance trend analysis in conjunction with the comprehensive rapping state set to obtain a cleaning efficiency optimization report.

[0008] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the pneumatic rapping operation data includes the rapper start / stop status, rapping air source pressure, number of consecutive rapping cycles, single rapping time window, temperature, and flue gas pressure difference.

[0009] The structural acoustic and vibration signals include vibration acceleration time history, strain time history, and vibration impact signals.

[0010] The preprocessing includes time alignment, noise reduction filtering, outlier removal, and classification.

[0011] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the high ash condition rapping dataset includes rapping control parameters, corresponding operating condition data, and acoustic and vibration characteristics.

[0012] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the rapping multimodal feature set includes working condition change features, acoustic vibration response features, and structural response features.

[0013] The specific steps for extracting time-frequency features from the high-gray-condition vibration dataset to obtain the vibration multimodal feature set are as follows:

[0014] The high-gray-condition vibration dataset is sliced ​​and recombined according to the single vibration time window to generate a multimodal vibration sample sequence.

[0015] A unified time series analysis and time-frequency feature extraction are performed on the multimodal rapping sample sequences to generate a multimodal rapping feature set.

[0016] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the rapping comprehensive state set includes a high ash index, a fatigue margin index, and a rapping energy consumption index.

[0017] The specific steps for generating the comprehensive vibration state set are as follows:

[0018] The high ash index is calculated based on the characteristics of working condition changes and acoustic and vibration response characteristics to quantify the degree of high ash in the dust removal area.

[0019] The fatigue margin index is calculated based on the structural response characteristics, and the structural fatigue margin is quantified by classifying the fatigue margin levels.

[0020] The vibration control and working condition information in the multimodal vibration sample sequence is extracted to calculate the vibration energy consumption index and quantify the vibration energy consumption.

[0021] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the specific steps for constructing the dual-constraint rapping optimization model are as follows:

[0022] The integrated state set of vibration is normalized and vectorized to form the vibration double-constraint state vector.

[0023] Based on the vibration dual-constraint state vector, the vibration comprehensive state set is jointly analyzed by the dual-constraint strength feature extraction algorithm to obtain the dual-constraint strength coefficient set.

[0024] Based on the set of dual-constraint strength coefficients and the comprehensive state set of vibration, vibration strength optimization rules are generated through dual-constraint collaborative mapping.

[0025] The dual-constraint rapping state vector is mapped and solidified with the rapping intensity optimization rule to construct a dual-constraint rapping optimization model.

[0026] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the generated region rapping optimization strategy refers to generating a region rapping optimization strategy by balancing the rapping intensity parameters through a dual-constraint rapping optimization model and a dual-objective trade-off optimization algorithm.

[0027] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the execution area rapping optimization strategy combines the high ash condition rapping dataset to identify the marginal cleaning effect of pneumatic rapping through acoustic vibration features. The specific steps are as follows:

[0028] The regional vibration optimization strategy is implemented, and a vibration event tagging sequence is generated by combining the high gray condition vibration dataset with regional strategy distribution and timestamp marking.

[0029] Based on the rapping event marker sequence, the acoustic vibration data segments of short time windows before and after the rapping are extracted by the acoustic vibration alignment time window truncation algorithm to obtain the acoustic vibration feature sequence;

[0030] Based on the acoustic and vibration characteristic sequence and the comprehensive state set of rapping, the marginal rapping effectiveness index is recorded by comparing the changes in acoustic and vibration response modes before and after rapping, and a marginal dust removal evaluation result set is generated.

[0031] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the step of obtaining the rapping execution result refers to statistically classifying effective rapping, redundant rapping, and over-rapping risk on the rapping event label sequence based on the marginal cleaning evaluation result set, and summarizing them to form the rapping execution result.

[0032] As a preferred embodiment of the pneumatic rapping cleaning efficiency optimization method under high ash conditions described in this invention, the steps are as follows: adaptively adjusting the dual-constraint rapping optimization model based on the rapping execution results, and combining the rapping comprehensive state set to perform cleaning performance trend analysis and obtain a cleaning efficiency optimization report.

[0033] Based on the results of the vibration execution, vibration performance deviation data are obtained through actual performance alignment analysis;

[0034] Extract the state deviation factors from the comprehensive state set of rapping, and combine them with the rapping efficiency deviation data to correct the set of dual-constraint strength coefficients, thereby obtaining the updated dual-constraint rapping optimization model;

[0035] By combining the updated dual-constraint rapping optimization model with the rapping execution results, the dust removal performance trend index is calculated, and a dust removal efficiency optimization report is obtained.

[0036] The beneficial effects of this invention are as follows: by constructing a comprehensive rapping state set that simultaneously characterizes high ash index, structural fatigue margin and rapping energy consumption, and by using acoustic vibration characteristics to identify the marginal cleaning efficiency of a single rapping and distinguish between effective rapping, redundant rapping and over-vibration risk, the invention achieves precise perception and closed-loop optimization control of the pneumatic rapping process under high ash conditions, thereby improving the cleaning efficiency of electrostatic precipitators under high ash conditions and reducing unit cleaning energy consumption. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 A flowchart illustrating a method for optimizing the efficiency of pneumatic rapping dust removal under high-ash conditions.

[0039] Figure 2 A flowchart for generating the integrated state set of rapping.

[0040] Figure 3A flowchart for generating a regional vibration optimization strategy.

[0041] Figure 4 A flowchart for obtaining a dust removal efficiency optimization report. Detailed Implementation

[0042] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0043] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0045] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for optimizing the efficiency of pneumatic rapping cleaning under high ash conditions, including the following steps:

[0046] S1: Collect pneumatic rapping operation data and structural acoustic and vibration signals, and obtain high-gray working condition rapping dataset through preprocessing.

[0047] S1.1: Pneumatic rapping operation data includes rapper start / stop status, rapping air source pressure, number of consecutive rapping cycles, single rapping time window, temperature, and flue gas pressure difference;

[0048] Specifically, the start and stop status of the rapper is collected by the switch signal in the rapper control device. For the rapping object, such as the dust collecting plate and discharge electrode of the electrostatic precipitator, and the dust removal equipment under high dust conditions, the control contact status at the start and end of each rapping is recorded as the start and stop status of the rapper.

[0049] The rapping air source pressure is acquired in real time by a pressure transmitter installed on the rapping air source pipeline. The continuous pressure measurement value output by the transmitter is recorded as the rapping air source pressure according to a fixed sampling period.

[0050] The number of consecutive rapping events is generated by counting the number of times the same rapper is continuously started and stopped within a set time period using a rapping control device.

[0051] The single vibration time window obtains the start and end time interval of each round of vibration by recording the timestamps of the vibration start signal and end signal issued by the vibration control device, and uses this time interval as the single vibration time window.

[0052] Temperature is collected by temperature sensors placed near the inlet and outlet flues of the electrostatic precipitator. The temperature values ​​measured by the temperature sensors within a single rapping time window are recorded as temperatures in chronological order.

[0053] The flue gas pressure difference is collected by differential pressure transmitters located upstream and downstream of the flue. The differential pressure measurement value output by the differential pressure transmitter is correlated with the corresponding single rapping time window and recorded as the flue gas pressure difference.

[0054] S1.2: Structural acoustic and vibration signals include vibration acceleration time history, strain time history, and vibration impact signals;

[0055] Specifically, the vibration acceleration time history is obtained by arranging vibration acceleration sensors at the connection points of the electrostatic precipitator shell or the dust collection electrode support beam, using a data acquisition device to continuously record the acceleration output at a uniform sampling frequency, and arranging them in chronological order to form a complete vibration acceleration time history.

[0056] The strain time history is formed by arranging resistance strain gauges at impact-prone locations on the dust collection plates, plate frame beams, or electrostatic precipitator shell, collecting voltage changes in the strain measurement bridge, and then storing them sequentially according to the sampling time.

[0057] The rapping impact signal is recorded by an impact sound pickup sensor or industrial microphone near the rapping hammer or on the outer surface of the housing. The sound pressure waveform generated at the moment of rapping is recorded at the same sampling frequency as the vibration acceleration sensor and strain gauge. The sound pressure waveform is then aligned with the start and end timestamps of the rapping to obtain the rapping impact signal record corresponding to a single rapping process.

[0058] S1.3: Preprocessing includes time alignment, noise reduction filtering, outlier removal, and classification.

[0059] Specifically, time alignment is completed based on timestamps using a single rapping time window as the boundary. Within each time window, noise filtering is performed on the rapping air source pressure and the number of consecutive rappings, and abnormal records are removed. The start / stop status of the rapper, rapping air source pressure, number of consecutive rappings, and single rapping time window after cleaning within the same time window are categorized and organized according to the time window number. The set of start / stop status, pressure, number of rappings, and time window length corresponding to each time window in the organized results are recorded as rapping control parameters.

[0060] Based on the start and end times of a single rapping time window, the timestamped measurements from the temperature sensor and differential pressure transmitter are time-aligned. Within each time window, the temperature and flue gas differential pressure sequences are smoothed, denoised, and abrupt change points are removed to eliminate abnormal measurements that deviate from the actual operating conditions. The cleaned temperature and flue gas differential pressure sequences within the same time window are then categorized and organized. The temperature and flue gas differential pressure records for each time window in the organized results generate corresponding operating condition data.

[0061] Based on the single rapping time window and the start and end timestamps of rapping, vibration acceleration segments, strain segments, and sound pressure waveform segments corresponding to the rapping process are extracted within each time window. Bandpass filtering is applied to suppress power frequency interference and background noise and remove high-amplitude impact segments. Then, the processed vibration acceleration time history, strain time history, and rapping impact signal are classified and organized according to the single rapping time window. The three types of acoustic and vibration data records that appear in groups within each time window in the organized results are used to obtain acoustic and vibration characteristics.

[0062] The vibration control parameters, corresponding operating condition data, and acoustic and vibration characteristics of a single vibration time window are aligned one by one according to a unified timestamp and stored as a single record to obtain the high-gray condition vibration dataset.

[0063] S2: Extract time-frequency features from the high-gray-condition vibration dataset to obtain a multimodal vibration feature set. Quantify the vibration performance state through a vibration state evaluation algorithm to generate a comprehensive vibration state set.

[0064] S2.1: The high-gray condition rapping dataset is sliced ​​and recombined according to the single rapping time window to generate a multimodal rapping sample sequence;

[0065] It should be noted that the rapping control parameters, corresponding operating condition data, and acoustic and vibration characteristics in the high-ash condition rapping dataset are read. All records are sorted and grouped according to the start and end times of the single rapping time window and the dust removal area in the rapping control parameters. Records with the same start and end times and the same dust removal area are grouped under the same single rapping time window identifier. Using each single rapping time window identifier as a slice unit, the rapping control parameters, corresponding operating condition data, and acoustic and vibration characteristics within the time range of the single rapping time window are extracted and arranged according to the time order and dust removal area to form a multimodal rapping sample containing the rapping control parameters, corresponding operating condition data, and acoustic and vibration characteristics. Multiple multimodal rapping samples are generated sequentially from front to back along the time axis according to the single rapping time window, and then concatenated sequentially according to the dust removal area and time order to generate a multimodal rapping sample sequence.

[0066] S2.2: Perform unified time series analysis and time-frequency feature extraction on the multimodal rapping sample sequences to generate a multimodal rapping feature set;

[0067] It should be noted that, with the start and end times of a single rapping time window as boundaries, interpolation processing with a uniform time step is performed on the temperature and flue gas pressure difference in the multimodal rapping sample sequence to ensure that the temperature and flue gas pressure difference fall on the same time axis; for the vibration acceleration time history, strain time history, and rapping impact signal in the multimodal rapping sample sequence, signal segments covering the single rapping time window are extracted.

[0068] Within a single vibration time window of each signal segment, the time series is converted into an envelope curve describing the change of vibration intensity over time, and the instantaneous amplitude is extracted. By accumulating and quantizing the envelope curve or instantaneous amplitude within a single vibration time window, an energy characteristic value reflecting the overall vibration intensity is obtained. The instantaneous amplitude and energy characteristic value of the same signal segment are used as acoustic vibration response characteristics.

[0069] By applying Fast Fourier Transform to the vibration acceleration time history, strain time history, and rapping impact signal within a single rapping time window, the time series is converted into a spectral distribution corresponding to frequency and amplitude; the spectral distribution of the same signal segment is used as a structural response feature.

[0070] Temperature and flue gas pressure difference are used as operating condition change characteristics, instantaneous amplitude and energy characteristic values ​​are used as acoustic and vibration response characteristics, and the spectral distribution of strain time history, rapping impact signal and vibration acceleration time history are used as structural response characteristics. The data are then grouped and stored according to the single rapping time window and the dust removal area as indexes to generate a rapping multimodal feature set.

[0071] S2.4: Calculate the high ash index based on the characteristics of working condition changes and acoustic vibration response to quantify the degree of high ash in the cleaning area;

[0072] It should be noted that quantitative features reflecting flue gas pressure difference and temperature are selected from the operating condition variation characteristics, while instantaneous amplitude and energy characteristic values ​​are selected from the acoustic and vibration response characteristics. Normalization is performed within the single rapping time window according to their respective maximum and minimum value ranges, transforming the operating condition variation characteristics and acoustic and vibration response characteristics into comparable dimensionless characteristics. The high ash index is then calculated, expressed as:

[0073] ;

[0074] in, It has a high gray index. To normalize the flue gas pressure difference, For normalized temperature, To normalize the instantaneous amplitude, This represents the normalized energy eigenvalue.

[0075] In the multimodal rapping sample sequence, the high gray index value is recorded for each cleaning area and stored together with the single rapping time window identifier and the cleaning area. The high gray index is used to quantify the degree of high gray in each cleaning area within different single rapping time windows. For example, within different single rapping time windows of the same cleaning area, a high gray index value of 0.8 indicates that the gray layer is thick and the dust accumulation is serious, while a high gray index value of 0.2 indicates that the gray layer is thin and the dust accumulation is light.

[0076] S2.5: Calculate the fatigue margin index based on the structural response characteristics, and quantify the structural fatigue margin by classifying fatigue margin levels;

[0077] It should be noted that, for each cleaning area and each single rapping time window, the strain time history in the structural response characteristics is traversed, and the maximum value of the strain time history is found as the peak strain; the difference between the maximum and minimum values ​​is taken as the strain change amplitude; the allowable strain range of the structural material corresponding to the cleaning area is read; based on the peak strain, strain change amplitude, and allowable strain range of the structural material, the fatigue margin index is calculated, expressed as:

[0078] ;

[0079] in, As a fatigue margin indicator, This refers to the upper limit of the allowable strain range for structural materials. The peak strain is the characteristic of the structural response. This represents the magnitude of strain change in the structural response characteristics.

[0080] Furthermore, the allowable strain range of structural materials is used to indicate the strain range within which heated surface tube banks, support beams, or shells can withstand long-term stress without yielding or irreversible fatigue damage under pneumatic vibration conditions. Generally, the lower limit is close to zero strain or measurement noise level, while the upper limit is determined as the elastic working limit based on the material's mechanical properties, wall thickness, service temperature, and engineering safety specifications. For example, the 12Cr1MoV alloy steel commonly used in boiler heated surfaces can have an allowable strain range of 0 to 0.002 under medium-temperature pneumatic vibration conditions, thus providing a reference boundary for assessing the structural fatigue margin based on peak strain and strain variation amplitude.

[0081] Multiple fatigue margin levels are defined based on the different ranges of the fatigue margin index values. For example, a fatigue margin index value above 0.8 is recorded as Level 1 fatigue margin, between 0.5 and 0.8 as Level 2 fatigue margin, and below 0.5 as Level 3 fatigue margin. The fatigue margin index values ​​and corresponding fatigue margin level labels are stored together in the rapping comprehensive status record, so that each dust removal area has a fatigue margin index and fatigue margin level for quantifying structural fatigue margin within each single rapping time window. For example, when the fatigue margin level is Level 1, moderately increased rapping intensity is allowed; when the fatigue margin level is Level 2, a balance is struck between high dust index and fatigue risk to avoid over- or under-rapping; when the fatigue margin level is Level 3, rapping intensity is limited.

[0082] S2.6: Extract rapping control and working condition information from multimodal rapping sample sequences, calculate rapping energy consumption index, and quantify rapping energy consumption;

[0083] It should be noted that, according to the single rapping time window and the dust removal area, the rapping control parameters corresponding to each single rapping time window, including the rapping air source pressure, single rapping time window, and number of consecutive rappings, are read from the multimodal rapping sample sequence, and multiple multimodal rapping samples are grouped into a statistical sequence. Then, within each single rapping time window, the ratio of the rapping air source pressure to the length of the single rapping time window is taken as the relative air consumption of a single rapping action. Within the same dust removal area, the sum of the relative air consumption of the number of consecutive rappings is taken as the rapping air consumption. Within the statistical sequence, the sum of all rapping air consumption in each consecutive rapping process according to the dust removal area is taken as the rapping energy consumption index.

[0084] The rapping energy consumption index, along with the corresponding dust removal area and statistical sequence identifier, is stored in the rapping comprehensive status record. The rapping energy consumption index is used to quantify the rapping energy consumption. For example, if the rapping energy consumption index of a certain dust removal area in a statistical sequence is higher than that of other dust removal areas, it can be determined that the rapping energy consumption of the dust removal area is too high in the statistical sequence. In subsequent rapping comprehensive status sets, priority should be given to optimizing and adjusting the rapping intensity parameters of the dust removal area.

[0085] The high ash index, fatigue margin index, and rapping energy consumption index are uniformly organized and combined according to the single rapping time window and the dust removal area identifier to generate a comprehensive rapping state set.

[0086] Furthermore, the rapping state assessment algorithm simultaneously quantifies the high ash index, fatigue margin index, and rapping energy consumption index within the same rapping time scale. This transforms the changes in high ash intensity, structural stress boundaries, and air source utilization during the ash removal process into a comparable comprehensive rapping state set. Consequently, it provides a unified and quantifiable state input for the subsequent dual-constraint rapping optimization model, avoiding over-vibration, under-vibration, or air waste caused by single-index decisions, and improving the precision and reliability of rapping control under high ash conditions.

[0087] S3: Based on the comprehensive state set of rapping, a dual-constraint rapping optimization model is constructed. By optimizing the configuration of rapping intensity parameters, a regional rapping optimization strategy is generated.

[0088] S3.1: Normalize and vectorize the integrated state set of rapping to form a rapping double-constraint state vector;

[0089] It should be noted that the range of high gray index and fatigue margin index corresponding to all dust removal areas and all single rapping time windows in the comprehensive rapping state set are statistically analyzed. The range of high gray index and fatigue margin index are used as reference intervals for subsequent normalized vectorization processing. For each state record in the comprehensive rapping state set, a linear normalization method is used to convert the high gray index into dimensionless high gray coordinates between zero and one, and the fatigue margin index into dimensionless fatigue coordinates between zero and one.

[0090] The larger the dimensionless high gray coordinate value, the more severe the high gray degree; the larger the dimensionless fatigue coordinate value, the more sufficient the structural safety margin. Combining the dimensionless high gray coordinate and the dimensionless fatigue coordinate generates a vibration double constraint state vector.

[0091] S3.2: Based on the vibration dual-constraint state vector, the vibration comprehensive state set is jointly analyzed by the dual-constraint strength feature extraction algorithm to obtain the dual-constraint strength coefficient set;

[0092] It should be noted that the vibration double-constraint state vector and the vibration time sequence number in the vibration comprehensive state set are extracted. Then, according to the vibration time sequence number, the dimensionless gray coordinate and dimensionless fatigue coordinate corresponding to each vibration cycle in the vibration double-constraint state vector are paired with the vibration state quantity of the same vibration cycle in the vibration comprehensive state set. Based on the two constraint components and the paired vibration state quantity in each pairing record, the double-constraint strength is calculated by the double-constraint strength feature extraction algorithm to generate the double-constraint strength feature vector. All double-constraint strength feature vectors are traversed in the order of vibration time sequence number and the double-constraint strength corresponding one-to-one with the vibration comprehensive state set is extracted. All double-constraint strengths are sorted and summarized to obtain the double-constraint strength coefficient set.

[0093] The expression for calculating the strength of a double constraint is:

[0094] ;

[0095] in, It is a double-constraint strength. Dimensionless gray coordinates For vibration energy consumption indicators, For dimensionless fatigue coordinates, This represents the fatigue safety margin ratio. This represents the energy consumption margin ratio for vibration.

[0096] Furthermore, the dual-constraint strength feature extraction algorithm combines dimensionless high-gray coordinates, dimensionless fatigue coordinates, high-gray index, fatigue margin index, and rapping energy consumption index within the same rapping cycle to extract dual-constraint strength coefficients that reflect the strength of high-gray cleaning requirements and the urgency of structural fatigue constraints. This compresses the originally dispersed and dimensionless multiple indicators into comparable dual-constraint strengths.

[0097] The superior method of calculating the dual-constraint strength facilitates the direct sorting of the areas by strength during the optimization of regional rapping, distinguishing the areas that need enhanced rapping and the areas that need suppressed rapping. On the other hand, it provides a unified constraint characterization basis for the subsequent dual-objective trade-off optimization algorithm, thereby reducing the intervention of human experience and improving the objectivity and precision of rapping strength configuration under high-ash conditions.

[0098] S3.3: Based on the set of dual-constraint strength coefficients and the comprehensive state set of rapping, rapping strength optimization rules are generated through dual-constraint collaborative mapping;

[0099] It should be noted that, in the set of dual-constraint strength coefficients, the dual-constraint strength coefficients corresponding to the rapping cycle are read according to the rapping time sequence number recorded in the rapping comprehensive state set; the dual-constraint strength coefficients are paired one by one with the dimensionless high gray coordinates and dimensionless fatigue coordinates recorded in the same rapping cycle in the rapping comprehensive state set to form a dual-constraint collaborative mapping input sample sequence.

[0100] Based on the input sample sequence of the dual-constraint co-mapping, the changes in the dual-constraint strength coefficient are compared with the changes in the dimensionless high gray coordinate and the dimensionless fatigue coordinate. During the comparison, a correspondence is formed between the changes in the dual-constraint strength coefficient and the changes in the dimensionless high gray coordinate and the dimensionless fatigue coordinate. For example, when the dual-constraint strength coefficient changes from 0.3 to 0.5, the dimensionless high gray coordinate changes from 0.4 to 0.6 and the dimensionless fatigue coordinate changes from 0.2 to 0.1. This set of synchronous changes is recorded as a set of correspondences between the changes in the dual-constraint strength coefficient and the changes in the dimensionless high gray coordinate and the dimensionless fatigue coordinate.

[0101] Based on the correspondence output by the dual-constraint collaborative mapping, the rapping strength adjustment direction and rapping strength adjustment range are divided according to the dual-constraint strength coefficient range and the dimensionless high gray coordinate and dimensionless fatigue coordinate range. The rapping strength optimization rules are generated by summarizing and organizing the range combinations.

[0102] S3.4: Map and solidify the dual-constraint rapping state vector with the rapping intensity optimization rules to construct a dual-constraint rapping optimization model;

[0103] It should be noted that, based on the rapping dual-constraint state vector and the rapping intensity optimization rules, a one-to-one correspondence is established between the rapping dual-constraint state vector components corresponding to each rapping time sequence number and the rapping intensity optimization rules. After establishing the one-to-one correspondence for all rapping time sequences, all rapping time sequences, rapping dual-constraint state vector components, and rapping intensity optimization rules are organized into a fixed-format mapping relationship set, and the mapping relationship set is solidified into a dual-constraint rapping optimization model that can directly output the rapping energy, rapping frequency, and rapping duration control quantities based on the rapping dual-constraint state vector.

[0104] Furthermore, the dual-constraint rapping optimization model integrates the rapping intensity optimization rules into a rule-based reasoning framework. This framework, based on the real-time input rapping dual-constraint state vector, searches for the optimal rule within the rapping intensity optimization rules and executes the control strategy within those rules, thereby outputting control values ​​for rapping energy, rapping frequency, and rapping duration.

[0105] S3.5: Based on the dual-constraint rapping optimization model, the rapping intensity parameters are configured by a dual-objective trade-off optimization algorithm to generate a regional rapping optimization strategy.

[0106] It should be noted that, according to the dust removal area, the rapping double-constraint state vector of each dust removal area in each rapping cycle is extracted from the rapping double-constraint state vector sequence, and each vector is input into the double-constraint rapping optimization model to obtain the corresponding candidate set of rapping strength parameters; the candidate set of rapping strength parameters, together with the high ash index and fatigue margin index corresponding to the same dust removal area and the same rapping cycle in the rapping comprehensive state set, are used as dual-objective evaluation inputs to calculate the dust removal effect score; for each dust removal area, the optimal combination of rapping strength parameters with the highest dust removal effect score is selected, and all optimal combinations of rapping strength parameters are organized into a rapping strength control scheme covering all dust removal areas according to the dust removal area index, generating a regional rapping optimization strategy.

[0107] The expression for calculating the dust removal effect score is:

[0108] ;

[0109] in, Rate the dust removal effect. For the remaining relative cleanliness, This is the total penalty factor.

[0110] Furthermore, the dual-objective trade-off optimization algorithm considers both the dust removal effect objective and the structural fatigue life or rapping energy consumption objective in the same optimization process. It puts the need for high dust index improvement, structural safety, and air source consumption into the same trade-off framework for comprehensive decision-making, instead of focusing solely on dust removal efficiency. This allows it to avoid extreme situations when configuring rapping strength parameters.

[0111] Better yet, the calculation of the dust removal effect score achieves coordinated optimization of the rapping pressure and number of consecutive rapping times in each dust removal area, and obtains a more balanced comprehensive operating state between dust removal effect, equipment life and operating energy consumption under high dust conditions. For example, the dust removal effect score comprehensively considers the reduction of high dust index, improvement of fatigue margin and changes in rapping energy consumption. The higher the score, the better the comprehensive effect between dust removal effect, structural safety and air consumption.

[0112] S4: Execute the regional rapping optimization strategy, combine the high-gray working condition rapping dataset to identify the edge cleaning effect of pneumatic rapping through acoustic vibration features, and obtain the rapping execution results.

[0113] S4.1: Execute the regional vibration optimization strategy, and generate a vibration event tag sequence by combining the high gray condition vibration dataset with regional strategy distribution and timestamp marking;

[0114] It should be noted that, according to the dust removal area and single rapping time window identifier, the rapping intensity parameters corresponding to each dust removal area in each single rapping time window are read from the regional rapping optimization strategy; the start / stop status, rapping air source pressure, number of consecutive rappings, and start / end time of the single rapping time window are read from the rapping control parameters corresponding to the same dust removal area and the same single rapping time window from the high-ash working condition rapping dataset; the rapping intensity parameters and rapping control parameters are matched one by one according to the dust removal area and single rapping time window identifier to generate regional strategy distribution records.

[0115] At the beginning of each single rapping time window, the rapping intensity parameters recorded in the area strategy are issued to control the rapping process in the dust removal area. At the start and end times of rapping, the corresponding timestamps are recorded. The dust removal area, the single rapping time window identifier, the start / stop status in the rapping control parameters, and the start and end timestamps of rapping are organized into a rapping event marker. All single rapping time windows are traversed sequentially along the time axis and all rapping event markers are added in turn to generate a rapping event marker sequence.

[0116] S4.2: Based on the vibration event marker sequence, the acoustic vibration data segments of short time windows before and after vibration are extracted by the acoustic vibration alignment time window truncation algorithm to obtain the acoustic vibration feature sequence;

[0117] It should be noted that, based on the start and end timestamps of each vibration event marker record in the vibration event marker sequence, continuous sampling points are selected on the corresponding vibration acceleration time history, strain time history, and vibration impact signal, respectively, within a fixed time interval before the start timestamp and after the end timestamp. Continuous sampling points within the same time range are spliced ​​into two acoustic vibration data segments, and classified into short-time window acoustic vibration data segments before and after vibration according to the time order of the vibration event marker sequence.

[0118] Furthermore, the acoustic-vibration alignment time window truncation algorithm, based on the rapping event marker, synchronously truncates short time segments before and after the rapping of the acoustic and vibration signals within the same single rapping time window, ensuring that the acoustic-vibration features strictly correspond to the same rapping impact process, eliminating the misalignment caused by time drift and asynchronous sampling, thereby obtaining more accurate and reliable acoustic-vibration response feature support.

[0119] Instantaneous acoustic vibration response features are extracted in chronological order within each short-time window acoustic vibration data segment before and after rapping. The acoustic vibration response features corresponding to each rapping event are then concatenated in the order they are marked on the time axis to obtain the acoustic vibration feature sequence.

[0120] S4.3: Based on the acoustic and vibration characteristic sequence and the comprehensive state set of rapping, the marginal rapping effectiveness index is recorded by comparing the changes in acoustic and vibration response modes before and after rapping, and a marginal dust removal evaluation result set is generated.

[0121] It should be noted that, based on the rapping time sequence number and dust removal area identifier recorded in the comprehensive rapping status, the acoustic and vibration response features before and after each rapping are extracted from the acoustic and vibration feature sequence and organized into one-to-one corresponding acoustic and vibration feature pairs.

[0122] For each pair of acoustic and vibration characteristics, the difference between the acoustic and vibration response characteristics of the short-time window after rapping and the acoustic and vibration response characteristics of the short-time window before rapping is used as the change in acoustic and vibration response mode. The change in acoustic and vibration response mode is combined with the dimensionless high gray coordinate change and dimensionless fatigue coordinate change recorded under the same rapping time number in the rapping comprehensive state set as the marginal rapping effectiveness index. All marginal rapping effectiveness indices and corresponding rapping comprehensive state records are sorted and summarized according to the rapping time number to generate the marginal dust removal evaluation result set.

[0123] S4.4: Based on the marginal dust removal evaluation result set, the effective rapping, redundant rapping and over-vibration risk are statistically classified on the rapping event label sequence, and the rapping execution results are summarized.

[0124] It should be noted that, based on the rapping time sequence and the dust removal area identifier, each marginal rapping efficiency index in the marginal dust removal evaluation result set is aligned with the corresponding rapping event in the rapping event label sequence to generate an extended sequence of rapping events with marginal rapping efficiency indices. When the marginal rapping efficiency index increases and the dimensionless high gray coordinate decreases, while the dimensionless fatigue coordinate remains unchanged, the rapping event is extended and recorded, marked as effective rapping. When the marginal rapping efficiency index decreases and the dimensionless high gray coordinate and dimensionless fatigue coordinate remain unchanged, it is marked as redundant rapping. When the marginal rapping efficiency index decreases and the dimensionless fatigue coordinate increases, it is marked as over-rapping risk. For example, when the marginal rapping efficiency index decreases by more than 10% and the dimensionless high gray coordinate changes by less than 0.05, it is marked as redundant rapping; when the dimensionless fatigue coordinate increases by more than 0.1, it is marked as over-rapping risk.

[0125] The results of all classification and judgment are statistically summarized according to the vibration time sequence number and the dust removal area identification to generate the vibration execution result.

[0126] S5: Adaptively adjust the dual-constraint rapping optimization model based on the rapping execution results, and conduct dust removal performance trend analysis in conjunction with the comprehensive rapping state set to obtain a dust removal efficiency optimization report.

[0127] S5.1: Based on the rapping execution results, obtain rapping performance deviation data through measured performance alignment analysis;

[0128] It should be noted that, based on the rapping execution results and the marginal cleaning evaluation results set, the rapping time sequence number and cleaning area identifier in the rapping execution results are extracted. The marginal rapping efficiency indicators with the same rapping time sequence number and the same cleaning area identifier are searched one by one in the marginal cleaning evaluation results set. The rapping execution results record and the corresponding marginal rapping efficiency indicator are merged into the measured rapping efficiency record.

[0129] Within each dust removal area, records marked as effective rapping results are selected from the measured rapping performance records. The marginal rapping performance index obtained from the selection is used as the baseline value of the marginal rapping performance of the dust removal area. The difference between the marginal rapping performance index corresponding to each rapping event in the same dust removal area and the baseline value of the marginal rapping performance of the dust removal area is used as the rapping performance deviation value. All rapping performance deviation values ​​are organized into a record sequence with rapping time sequence number, dust removal area identifier and rapping performance deviation value according to the rapping time sequence number and dust removal area identifier, thus generating rapping performance deviation data.

[0130] S5.2: Extract the state deviation factor from the comprehensive state set of rapping, and combine it with the rapping efficiency deviation data to correct the set of double constraint strength coefficients, and obtain the updated double constraint rapping optimization model;

[0131] It should be noted that, based on the rapping time sequence and the dust removal area identifier, each record in the rapping comprehensive state set is compared with the previous record in the same area. The increase or decrease of the dimensionless high gray coordinate, the increase or decrease of the dimensionless fatigue coordinate, and the increase or decrease of the rapping energy consumption index are used as the state deviation factor for this record (composed of the changes in the dimensionless high gray coordinate, dimensionless fatigue coordinate, and rapping energy consumption index in the rapping comprehensive state set and the rapping double constraint state vector). The state deviation factor is matched with the rapping performance deviation data and the double constraint strength coefficient set one by one according to the rapping time sequence and the dust removal area identifier. When the state deviation factor is manifested as an increase in the dimensionless high gray coordinate, the double constraint strength coefficient of the corresponding rapping cycle is decreased; when the state deviation factor is manifested as a low rapping energy consumption index, the double constraint strength coefficient of the corresponding rapping cycle is increased; when the state deviation factor is manifested as an increase in the dimensionless fatigue coordinate, the double constraint strength coefficient of the corresponding rapping cycle is decreased; when the rapping performance deviation data is stable and the state deviation factor fluctuates little, the double constraint strength coefficient is kept unchanged.

[0132] After traversing all double-constraint strength coefficients according to the rapping time sequence, a new set of double-constraint strength coefficients is formed. The updated set of double-constraint strength coefficients is then used to replace the set of double-constraint strength coefficients in the double-constraint rapping optimization model to obtain the updated double-constraint rapping optimization model.

[0133] S5.3: Combine the updated dual-constraint rapping optimization model with the rapping execution results to calculate the dust removal performance trend index and obtain a dust removal efficiency optimization report.

[0134] It should be noted that, based on the updated dual-constraint rapping optimization model, rapping execution results, and rapping comprehensive state set, the rapping dual-constraint state vector, dimensionless high-gray coordinates, dimensionless fatigue coordinates, and rapping energy consumption index corresponding to each rapping are extracted from the rapping comprehensive state set according to the rapping time sequence and dust removal area identifier; the rapping dual-constraint state vectors are then input into the updated dual-constraint rapping optimization model one by one to obtain the recommended dual-constraint strength parameter set corresponding to each rapping.

[0135] According to the rapping time sequence and the dust removal area identifier, the recommended dual-constraint strength parameter group is matched with the effective rapping mark, redundant rapping mark and over-vibration risk mark in the rapping execution result. The changes of dimensionless high ash coordinate, dimensionless fatigue coordinate and rapping energy consumption index on the time axis are respectively used as the dust removal performance trend indicators of the high ash level change trend, structural fatigue risk change trend and rapping energy consumption change trend.

[0136] Based on the dust removal performance trend indicators, rapping execution results, and changes in recommended dual constraint strength parameters of the dust removal area, a dust removal efficiency optimization report is obtained.

[0137] In summary, this invention achieves precise perception and closed-loop optimization control of the pneumatic rapping process under high ash conditions by constructing a comprehensive rapping state set that simultaneously characterizes high ash index, structural fatigue margin, and rapping energy consumption, and by utilizing acoustic vibration characteristics to identify the marginal ash removal efficiency of a single rapping and distinguish between effective rapping, redundant rapping, and over-vibration risk. This improves the ash removal efficiency of electrostatic precipitators under high ash conditions and reduces unit ash removal energy consumption.

[0138] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing the pneumatic rapping ash efficiency under high-ash conditions, characterized by: The application relates to a method for constructing a double-constraint vibration-aided dust removal optimization model. The vibration-aided dust removal operation data and structure vibration signals are collected and preprocessed to obtain a high-ash condition vibration-aided dust removal data set; The high-ash condition vibration-aided dust removal data set is subjected to time window time-frequency feature extraction to obtain a vibration-aided dust removal multi-modal feature set, and the specific steps are as follows, The vibration-aided dust removal multi-modal feature set comprises working condition change features, vibration response features and structure response features; The high-ash condition vibration-aided dust removal data set is sliced and reorganized according to a single vibration time window to generate a vibration-aided dust removal multi-modal sample sequence; The vibration-aided dust removal multi-modal sample sequence is subjected to unified time sequence analysis and time-frequency feature extraction to generate a vibration-aided dust removal multi-modal feature set; The vibration-aided dust removal performance state is quantified through a vibration-aided dust removal state evaluation algorithm to generate a vibration-aided dust removal comprehensive state set, and the specific steps are as follows, The vibration-aided dust removal comprehensive state set comprises a high-ash index, a fatigue margin index and a vibration-aided dust removal energy consumption index; The high-ash index is calculated according to the working condition change features and the vibration response features to quantify the high-ash degree of the dust removal area; The fatigue margin index is calculated according to the structure response features to quantify the structure fatigue margin by dividing the fatigue margin level; The vibration-aided dust removal energy consumption index is calculated according to the vibration-aided dust removal control and working condition information in the vibration-aided dust removal multi-modal sample sequence to quantify the vibration-aided dust removal energy consumption; Based on the vibration-aided dust removal comprehensive state set, a double-constraint vibration-aided dust removal optimization model is constructed, vibration-aided dust removal intensity parameters are optimized and configured through the double-constraint vibration-aided dust removal optimization model, and a regional vibration-aided dust removal optimization strategy is generated; The regional vibration-aided dust removal optimization strategy is executed, the marginal dust removal effect of the vibration-aided dust removal is identified through vibration features in combination with the high-ash condition vibration-aided dust removal data set, and a vibration-aided dust removal execution result is obtained; The double-constraint vibration-aided dust removal optimization model is adaptively adjusted according to the vibration-aided dust removal execution result, dust removal performance trend analysis is carried out in combination with the vibration-aided dust removal comprehensive state set, and a dust removal efficiency optimization report is obtained.

2. The method according to claim 1, wherein the method is characterized in that: The vibration-aided dust removal operation data comprises vibration-aided dust remover start-stop states, vibration-aided dust removal gas source pressures, continuous vibration-aided dust removal times, single vibration-aided dust removal time windows, temperatures and flue gas pressure differences. The structure vibration signals comprise vibration acceleration time histories, strain time histories and vibration-aided dust removal knocking signals. The preprocessing comprises time alignment, denoising filtering, abnormal value elimination and classification and arrangement.

3. The method according to claim 2, wherein the method is characterized in that: The high-ash condition vibration-aided dust removal data set comprises vibration-aided dust removal control parameters, corresponding operation condition data and vibration features.

4. The method according to claim 3, wherein the method is characterized in that: The double-constraint vibration-aided dust removal optimization model is constructed, and the specific steps are as follows, The vibration-aided dust removal comprehensive state set is subjected to normalized vectorization processing to form a vibration-aided dust removal double-constraint state vector; Based on the vibration-aided dust removal double-constraint state vector, the vibration-aided dust removal comprehensive state set is subjected to joint analysis through a double-constraint intensity feature extraction algorithm to obtain a double-constraint intensity coefficient set; Based on the double-constraint intensity coefficient set and the vibration-aided dust removal comprehensive state set, a vibration-aided dust removal intensity optimization rule is generated through double-constraint collaborative mapping; The vibration-aided dust removal double-constraint state vector and the vibration-aided dust removal intensity optimization rule are mapped and solidified to construct the double-constraint vibration-aided dust removal optimization model.

5. The method according to claim 4, wherein the method is characterized in that: The regional vibration-aided dust removal optimization strategy is generated by weighting and configuring the vibration-aided dust removal intensity parameters through a double-target weighting optimization algorithm based on the double-constraint vibration-aided dust removal optimization model.

6. The method according to claim 5, wherein the method is characterized in that: The marginal dust removal effect of the vibration-aided dust removal is identified through vibration features in combination with the high-ash condition vibration-aided dust removal data set, and the specific steps are as follows, The area shaking optimization strategy is executed, and a shaking event marker sequence is generated by combining a high-ash working condition shaking data set, a region strategy, and a time stamp marker; Based on the shaking event marker sequence, the sound and vibration data segments before and after shaking are extracted by a sound and vibration alignment time window interception algorithm, and a sound and vibration feature sequence is obtained; According to the sound and vibration feature sequence and the shaking comprehensive state set, the marginal shaking efficiency index is recorded by comparing the sound and vibration response mode change before and after shaking, and a marginal dust removal evaluation result set is generated.

7. The method according to claim 6, wherein the method is characterized in that: The acquisition of the shaking execution result refers to the statistical classification of effective shaking, redundant shaking, and over-shaking risk based on the marginal dust removal evaluation result set on the shaking event marker sequence, and the shaking execution result is formed by summarizing.

8. The method according to claim 7, wherein the method is characterized in that: According to the shaking execution result, the double-constraint shaking optimization model is adaptively adjusted, and the dust removal performance trend is analyzed in combination with the shaking comprehensive state set to obtain a dust removal efficiency optimization report, and the specific steps are as follows, Based on the shaking execution result, the shaking efficiency deviation data is obtained by aligning the measured efficiency; The state deviation factor in the shaking comprehensive state set is extracted, and the double-constraint strength coefficient set is corrected in combination with the shaking efficiency deviation data to obtain an updated double-constraint shaking optimization model; In combination with the updated double-constraint shaking optimization model and the shaking execution result, the dust removal performance trend index is calculated to obtain a dust removal efficiency optimization report.

Images