Fly ash separation fineness adjusting method based on negative pressure feedback

Abstract

The present application belongs to the technical field of wind force winnowing, and discloses a fly ash separation fineness adjusting method based on negative pressure feedback, a multi-dimensional synchronous data acquisition system is built, four types of core data of equipment operation, material working condition, airflow field state and finished product fine powder quality are collected, abnormal interference is eliminated and standardization processing is completed through self-adaptive preprocessing technology; the system can quantitatively calculate the fineness deviation amplitude and the change rate, automatically determine the separation working condition grade, split the two types of deviation sources of material property fluctuation and equipment airflow field parameter drift, and then through the hierarchical incremental negative pressure regulation and control and the three-zone collaborative flow field adaptation of the separation cavity, the particle classification critical particle size is controlled, the problems of fine powder carrying coarse particles and qualified fine powder being intercepted by coarse powder are avoided; through the real-time comparison of the finished product fineness and the standard interval by the industrial control system, the quantitative deviation value and the change trend are quantified, and the working condition grade is automatically determined combined with the preset correlation model, and the core trigger source of separation abnormality is quickly located.

Description

Technical Field

[0001] This invention belongs to the field of wind power classification technology, specifically a method for adjusting the fineness of fly ash separation based on negative pressure feedback. Background Technology

[0002] Fly ash, as a core raw material for the resource utilization of solid waste from the thermal power industry, needs to be classified into coarse and fine powders through wind separation. The fineness of the finished product directly determines the activity, dosage, and quality of the final building materials. Wind-driven negative pressure separation is currently the mainstream dry process for large-scale fly ash separation. Existing fly ash negative pressure separation technologies generally adopt an open-loop fixed parameter control mode, that is, after the equipment is debugged, the core parameters such as the negative pressure of the separation chamber, the induced draft air volume, and the classification speed are kept constant for a long time. Only passive fine-tuning is performed by manually sampling the fineness of the finished product at regular intervals. This has the following technical problems:

[0003] The dynamic fluctuations of the incoming fly ash conditions are large. Real-time changes in the combustion conditions of raw coal, the moisture content of raw ash, the feed flow rate, and the original particle size distribution can cause the sorting threshold of the fixed negative pressure airflow field to shift, resulting in problems such as fine powder being entrained with coarse particles and qualified fine powder being intercepted by coarse powder. The fineness stability of the finished product is poor and the pass rate fluctuates greatly.

[0004] Manual adjustments are extremely slow and cannot meet the needs of continuous operation of the production line. Frequent manual intervention can easily cause production line downtime and disordered operation, significantly reducing sorting efficiency.

[0005] Traditional regulation involves step-like large-scale parameter adjustments without a fine-grained feedback correction mechanism. This easily leads to two extreme problems: excessively high negative pressure causing a surge in energy consumption, and excessively low negative pressure causing sorting failure, resulting in low energy utilization.

[0006] Existing technologies only focus on adjusting a single negative pressure parameter, without establishing a linkage feedback model between negative pressure and finished product fineness and working condition parameters. They lack adaptive compensation capabilities and cannot adapt to complex sorting scenarios with multiple working conditions. Summary of the Invention

[0007] The purpose of this invention is to provide a method for adjusting the fineness of fly ash sorting based on negative pressure feedback, so as to solve one or more problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for adjusting the fineness of fly ash sorting based on negative pressure feedback, comprising the following specific steps:

[0009] Furthermore, in the working condition modeling stage, a hierarchical multi-working condition coupling correlation model is constructed. The correlation model has the functions of working condition hierarchical threshold calibration, particle size critical value coupling correction, and sorting purity weight compensation. The core working condition range of fly ash wind-powered negative pressure sorting is locked, and three types of fluctuating working conditions are selected: gradient feed flow rate, raw ash moisture content, and original particle size distribution. Sorting parameters such as airflow field wind speed and swirling airflow intensity are matched synchronously, and the optimal parameters for steady-state operation are calibrated. The parameter types include five types of benchmark data: sorting chamber reference negative pressure, staged airflow velocity, 45μm sieve residue fineness, coarse powder fineness content, and fine powder entrainment rate.

[0010] Based on calibration data, a three-dimensional nonlinear correlation model is fitted for the negative pressure threshold, fineness qualification rate, and sorting purity corresponding to light, medium, and heavy working conditions. The standardized fineness deviation threshold range is divided and bound to the airflow field working conditions to formulate a stratified judgment standard. The correlation model has a built-in dynamic correction coefficient for particle size critical value, which can complete the adaptive compensation of negative pressure threshold according to the real-time change of the original fly ash particle size.

[0011] Furthermore, the parameter acquisition stage is based on the correlation model and working condition judgment criteria constructed in the working condition modeling stage. It sets the data acquisition frequency, builds an airflow sorting sensor acquisition system, and synchronously collects four types of core data: equipment operation, material working condition, airflow field, and finished product indicators. These include parameters such as airflow velocity in the three zones of the sorting chamber, return air pressure difference, classifier wheel airflow torque, real-time negative pressure value of the sorting chamber, induced draft fan airflow, classifier wheel speed, instantaneous feed flow rate, raw ash moisture content, feed particle size distribution, real-time sieve residue value of finished fine powder, and fineness content of coarse powder.

[0012] An adaptive filtering and noise reduction algorithm is used to preprocess the collected data, removing abnormal interference data caused by instantaneous material impact, airflow turbulence disturbance, and equipment vibration during the wind sorting process. The effective data after preprocessing is then standardized and uploaded to the industrial control system.

[0013] The abnormal data removal adopts a continuous periodic comparison rule, which performs a horizontal comparison of multiple sets of continuously collected data from a single sensor, and removes only the single set of abnormal values ​​with instantaneous changes, while retaining stable and valid data in adjacent periods. For continuously occurring abnormal data, the system will mark the sensor location and prompt the data abnormality, and at the same time enable the backup data fitting logic to perform short-term fitting supplement based on historical stable data and related parameters.

[0014] Furthermore, the deviation tracing stage is based on the standardized operation data preprocessed in the parameter acquisition stage. The industrial control system compares the fineness data of the finished fine powder with the standard sieve residue range of primary and secondary fly ash in real time, quantifies and calculates the fineness deviation value and the deviation change rate, and completes the automatic determination of the working condition level in combination with the correlation model.

[0015] The system binds material operating condition fluctuation data and airflow field parameter drift data to distinguish the two types of causes of fineness fluctuations. The two types of causes are material property fluctuations caused by changes in moisture content, particle size, and feed flow rate, and airflow separation field equipment parameter drifts caused by negative pressure offset, flow field unevenness, and classifier wheel speed deviation. This enables the separation of abnormal equipment failures from normal material operating condition fluctuations and pinpoints the root cause of control triggers.

[0016] The deviation tracing process is executed cyclically according to a fixed cycle. Within each execution cycle, standardized data from multiple consecutive collection cycles are called. The data call range includes the current cycle and historical data from the previous few cycles. During the tracing process, single-cycle data comparison is completed first, followed by multi-cycle data trend fitting. All data calls and calculations are executed in a preset time sequence.

[0017] Furthermore, the negative pressure control stage is based on the working condition level and the cause of fineness fluctuation determined in the deviation tracing stage. It adopts a graded incremental PID feedback adjustment algorithm to complete the dynamic correction of the sorting negative pressure. According to the working condition level, the corresponding parameter adjustment logic is matched to realize the sorting negative pressure calibration.

[0018] Under slight fluctuation conditions, the incremental correction mode is adopted to control the frequency adjustment of the induced draft fan within the preset range and maintain the airflow field balance in the sorting chamber. Under moderate deviation conditions, the negative pressure compensation is calculated through the layered coupling model, and the incremental adjustment step size is adaptively matched in combination with real-time feed flow rate and raw ash moisture content data to smoothly calibrate the negative pressure threshold.

[0019] Under severe abnormal operating conditions, a gradient progressive incremental adjustment mode is adopted, and the operating condition early warning linkage mechanism is activated simultaneously. The negative pressure parameters are corrected in stages and layers to stabilize the airflow field state. The algorithm incremental step size can be adaptively and dynamically adjusted according to the operating condition deviation level and the real-time state of the airflow field.

[0020] The execution cycle of the graded incremental PID control is set to a fixed duration synchronized with the material conveying sequence of the production line. After a single control command is issued and executed, the system pauses calculation and waits for a preset duration until the airflow inside the sorting chamber completes state reconstruction before starting the next round of data acquisition and parameter calculation. The command execution within the control cycle adopts a single-step completion mode. Each level of negative pressure compensation is executed completely according to the algorithm output result without interruption or splitting. A fixed time interval is set between two adjacent control calculations, and the interval duration is preset according to the sorting chamber volume and airflow response speed.

[0021] After the negative pressure correction is completed, the system continuously collects pressure and flow rate data in the rising airflow zone, swirling classification zone, and settling discharge zone of the sorting chamber. The collection period is consistent with the sensor sampling frequency. The system performs difference calculation on the monitoring data of the same time sequence in each zone and continuously records the data change amplitude in multiple collection cycles. The system compares the real-time fluctuation value of the data in each zone with the preset static threshold range and calculates the data change between adjacent cycles. When the data fluctuation value of all monitoring points falls within the preset range and the data change between adjacent cycles remains within the limited range, the system determines that the flow field has reached a stable state. The confirmation of the stable state of the flow field needs to be verified by comparing data from multiple consecutive cycles. The verification process is executed according to a fixed time sequence and calculation logic.

[0022] Furthermore, the flow field adaptation stage is based on the dynamic correction results of the negative pressure of the sorting completed in the negative pressure control stage. Through the principles of wind-driven airflow sorting and centrifugal classification, a three-zone coordinated control system of rising airflow zone, swirling classification zone, and settling discharge zone is established to coordinately adapt and control the flow field parameters of the three functional areas of the sorting chamber.

[0023] For the rising airflow zone, the opening of the air inlet distribution plate is dynamically adjusted; for the swirling classification zone, the auxiliary speed of the classification wheel is synchronously calibrated to stabilize the centrifugal separation airflow field and lock the critical particle size for particle classification; for the settling discharge zone, the airflow ratio of the return air duct is adaptively adjusted to balance the pressure of the settling airflow.

[0024] During the coordinated control of the rising airflow zone, the swirling classification zone, and the settling discharge zone, the parameter adjustments of each functional zone are set with independent single adjustment range limits. The opening of the air inlet distribution plate is adjusted in fixed angle steps, the speed of the classifier wheel is adjusted gradually according to a fixed speed gradient, and the air volume ratio of the return air duct is adjusted in segments according to a fixed ratio. The adjustment range limits of the three zones are all pre-calibrated based on the steady-state constraints of the airflow field. The parameter adjustment of any zone must wait for the adjustment action of the previous zone to be completed before it can be started.

[0025] The activation of the three-zone coordinated control requires two prerequisites: first, the negative pressure correction process must be completed; and second, the flow field in the sorting chamber must reach a preset stable state. Once both conditions are met, the system will automatically trigger the control command. If either condition is not met, the activation process will be paused, and subsequent operations will continue until the condition is met. The control activation command is automatically issued by the industrial control system according to fixed logic.

[0026] Furthermore, the closed-loop verification stage is based on the flow field collaborative adaptation and control operation completed in the flow field adaptation stage. It establishes a three-level control mechanism of adjustment effect verification, system error compensation, and secondary correction. It continuously collects finished product fineness, coarse powder fineness content, and airflow uniformity data for the next three operation cycles to evaluate the parameter adjustment effect.

[0027] When the fineness of the finished product returns to the standard range, the current operating parameters and flow field ratio parameters are locked. When there is a small residual deviation in the fineness of the finished product, the system calculates the adjustment error coefficient and performs pre-compensation on the subsequent PID incremental adjustment parameters. When the adjustment effect does not meet the standard requirements, the system automatically backtracks the material operating conditions, airflow field parameters, and negative pressure operation data to re-determine the cause of the deviation and start secondary graded adjustment.

[0028] Furthermore, the iterative learning phase builds a dynamic self-learning database based on the adjustment data, error correction data, and operating condition record data from the closed-loop verification phase. The dynamic self-learning database has the functions of operating condition classification iteration, long-term slow variable compensation, and automatic parameter calibration. The system classifies and stores the optimal negative pressure parameters, three-zone flow field ratio parameters, fineness compliance data, and deviation traceability records under different material operating conditions and different airflow field states to construct an operating condition parameter database.

[0029] By combining long-term operating variables such as seasonal humidity changes in the production line, raw coal quality fluctuations, and long-term equipment wear, the hierarchical negative pressure and fineness correlation model is optimized through big data iteration. The benchmark negative pressure threshold, PID increment step size, and three-zone flow field ratio corresponding to each operating condition are updated simultaneously. The system can adaptively match short-term instantaneous operating condition fluctuations with long-term seasonal and equipment aging operating condition changes.

[0030] After the model iteration is completed, the system first stores the newly generated parameters in the spare partition of the dynamic self-learning database, and then executes the parameter verification process. After the verification is successful, the parameters in the spare partition are switched to the running partition to replace the original control parameters. Once the new parameters take effect, they immediately participate in the next round of working condition judgment and adjustment calculation.

[0031] The beneficial effects of this invention are as follows:

[0032] 1. This invention establishes a multi-dimensional synchronous data acquisition system, collecting four core data categories: equipment operation, material conditions, airflow field status, and finished fine powder quality. Adaptive preprocessing technology is used to eliminate abnormal interference and complete standardization. The system can quantitatively calculate the amplitude and rate of change of fineness deviation, automatically determine the sorting condition level, and separate the two root causes of deviation: material property fluctuations and equipment airflow field parameter drift. Furthermore, through graded incremental negative pressure control and three-zone coordinated flow field adaptation in the sorting chamber, the critical particle size for particle grading is controlled, preventing fine powder from carrying coarse particles and qualified fine powder from being trapped by coarse powder. This ensures that the finished product fineness meets standards and improves the purity and quality of fly ash sorting.

[0033] 2. This invention uses an industrial control system to compare the fineness of finished products with the standard range in real time, quantifies the deviation value and trend of change, and automatically completes the determination of the working condition level by combining a preset correlation model, quickly locating the core triggering root cause of sorting anomalies; it adopts a graded incremental PID feedback adjustment algorithm, matching a dedicated progressive negative pressure correction strategy for different working condition levels, and is equipped with a three-level closed-loop verification mechanism of adjustment effect verification, system error compensation, and secondary correction, continuously monitoring the control effect and performing parameter locking, error pre-compensation, or secondary graded adjustment; the entire process adopts a smooth and progressive parameter adjustment method, avoiding airflow field turbulence and working condition oscillation caused by step adjustment, and reducing manual operation and maintenance costs.

[0034] 3. This invention employs a graded incremental PID feedback control algorithm, which can adaptively adjust the negative pressure control step size based on the level of operating condition deviation and the real-time airflow field state, calibrating the negative pressure parameters of the sorting chamber and avoiding energy consumption surges or sorting failures caused by abnormal negative pressure. By establishing a three-zone collaborative control system for the sorting chamber, the parameters are adjusted differently for the airflow characteristics of different functional areas to balance the overall airflow field distribution. At the same time, a dynamic self-learning database is constructed to iteratively optimize the operating condition correlation model and core control parameters based on operating condition data. This allows for simultaneous adaptation to short-term instantaneous operating condition fluctuations and the effects of slow variables such as long-term equipment aging and environmental changes, continuously updating the optimal benchmark parameters and control strategies under various operating conditions, thus extending the service life of the equipment. Attached Figure Description

[0035] Figure 1 This is a flowchart illustrating the overall process for adjusting the fineness of fly ash sorting based on negative pressure feedback according to the present invention.

[0036] Figure 2 This is a flowchart of the graded incremental PID negative pressure adaptive regulator of the present invention. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] like Figures 1 to 2 As shown, this embodiment of the invention provides a method for adjusting the fineness of fly ash sorting based on negative pressure feedback, including the following specific steps:

[0039] In this embodiment of the invention, a hierarchical multi-condition coupling correlation model is constructed in the working condition modeling stage. The correlation model has the functions of working condition hierarchical threshold calibration, particle size critical value coupling correction, and sorting purity weight compensation. The core working condition range of fly ash wind-driven negative pressure sorting is locked, and three types of fluctuating working conditions are selected: gradient feed flow rate, raw ash moisture content, and original particle size distribution. The gradient feed flow rate is 20–80 t / h, the raw ash moisture content is 0.5%–3.5%, and the original particle size reference range is 20–100 μm.

[0040] Synchronously match sorting parameters such as airflow field wind speed and swirling airflow intensity, and calibrate the optimal parameters for steady-state operation group by group. The parameter types include five types of benchmark data: sorting chamber reference negative pressure, graded airflow velocity, 45μm sieve residue fineness, coarse powder fineness content, and fine powder entrainment rate.

[0041] Steady-state operating baseline data was obtained using a gradient operating condition traversal calibration method. Multiple sets of gradient operating condition points were set within the full fluctuation range of feed flow rate, raw ash moisture content, and original particle size distribution. Each set of operating condition points was kept running continuously and stably for no less than 30 minutes. Multiple sets of parallel data were collected simultaneously, and discrete values ​​were removed. The average value was taken as the steady-state baseline data under that operating condition. The operating condition stratification threshold was determined through statistical analysis of a large amount of calibration data and comprehensively optimized in combination with indicators such as fineness qualification rate, sorting energy consumption, and equipment load during long-term operation of the production line.

[0042] Based on a large amount of calibration data, a three-dimensional nonlinear correlation model was fitted for the negative pressure threshold, fineness qualification rate, and sorting purity corresponding to light, medium, and heavy working conditions. The standardized fineness deviation threshold range was divided and bound to the airflow field working conditions. A stratified judgment standard was formulated: fineness deviation ≤3% is a slight airflow fluctuation working condition, 3% < deviation ≤8% is a medium working condition deviation working condition, and deviation >8% is a severe airflow turbulence abnormal working condition.

[0043] The three-dimensional nonlinear correlation model of negative pressure threshold, fineness qualification rate and sorting purity corresponding to light, medium and heavy working conditions is constructed by a third-order polynomial regularized fitting algorithm. During the fitting process, the L2 regularization factor is introduced to effectively suppress the risk of model overfitting.

[0044] The model uses a fineness deviation of 3% and 8% as the critical threshold for classification under working conditions, and divides the sorting scenario into three states: stable operation, slight deviation, and severe disorder. The built-in dynamic correction coefficient for the critical particle size has a value range of 0.8~1.2. It can perform online adaptive calculation based on the real-time particle size distribution of the original fly ash, and correct the corresponding mapping relationship between the negative pressure threshold and the critical particle size for classification in real time, so that the model can operate in industrial sites where the particle size, moisture content, and flow rate of the incoming material are constantly fluctuating.

[0045] Formula for dynamic correction coefficient of critical particle size:

[0046]

[0047] This represents the dynamic correction coefficient for the critical particle size, used to correct the negative pressure threshold for sorting in real time and adapt to the real-time changes in the particle size of native fly ash.

[0048] This represents the real-time average particle size of the raw fly ash, which is the average particle size value of fly ash obtained by real-time detection during the feeding process.

[0049] This represents the average particle size of the fly ash sorting benchmark, which is the average size value of the standard fly ash particles calibrated in advance by the system.

[0050] The correlation model has a built-in dynamic correction coefficient for the critical particle size, which can adaptively compensate for the negative pressure threshold according to the real-time changes in the particle size of the original fly ash, thereby suppressing the entrainment of fine powder and the retention of coarse powder.

[0051] The hierarchical multi-condition coupling association model adopts a three-layer feedforward deep neural network architecture customized for the fly ash wind separation scenario. The layers adopt a fully connected weighted transmission mode to complete the deep extraction and coupling calculation of the condition features. The input layer is set with 5 sets of standardized input nodes, which are respectively connected to five types of real-time sensing data: feed flow rate, raw ash moisture content, original particle size distribution, airflow field wind speed, and swirling airflow intensity.

[0052] The hidden layer consists of two levels of hidden layers. The first hidden layer is responsible for feature encoding, normalization and noise reduction fusion of multi-source working condition data. The second hidden layer integrates three core functional units: working condition layer threshold calculation, particle size critical value coupling correction and sorting purity weight compensation, to realize bidirectional collaborative mapping between material properties and airflow field parameters.

[0053] The output layer outputs three core control parameters: negative pressure threshold, fineness qualification rate, and sorting purity. The model training employs an adaptive mini-batch gradient descent optimization algorithm, using 1200 sets of steady-state calibration data covering the entire operating range of fly ash sorting as training samples and 300 sets of independent operating condition data as validation samples. The initial learning rate is set to 0.01, the batch size to 32, and the maximum number of iterations to 600. Convergence criteria are set at an absolute error in fineness prediction ≤1% and a model fit R² ≥ 0.98. After training, the model can respond in real-time to fluctuations in incoming material and changes in the airflow field.

[0054] The dynamic correction coefficient for the critical particle size adopts a real-time acquisition, real-time calculation, and real-time update execution mode. The system acquires the original fly ash particle size distribution data at fixed acquisition intervals, substitutes the data into the preset parameter mapping process to complete the coefficient value calculation, and directly replaces the correction coefficient currently used in the model after the calculation is completed. After the coefficient is updated, it immediately participates in the subsequent negative pressure threshold calculation process.

[0055] In this embodiment of the invention, the parameter acquisition stage is based on the correlation model and working condition judgment criteria constructed in the working condition modeling stage. A data acquisition frequency of 100ms / time is set, and an airflow sorting sensor acquisition system is built to simultaneously collect four types of core data: equipment operation, material working condition, airflow field, and finished product indicators. These include parameters such as airflow velocity in the three zones of the sorting chamber, return air pressure difference, classifier wheel airflow torque, real-time negative pressure value of the sorting chamber, induced draft fan airflow, classifier wheel speed, instantaneous feed flow rate, raw ash moisture content, feed particle size distribution, real-time sieve residue value of finished fine powder, and fineness content of coarse powder.

[0056] An adaptive filtering and noise reduction algorithm is used to preprocess the collected data, removing abnormal interference data caused by instantaneous material impact, airflow turbulence disturbance, and equipment vibration during the wind sorting process. The effective data after preprocessing is then standardized and uploaded to the industrial control system.

[0057] The adaptive filtering and noise reduction algorithm is a variable step size least mean square (LMS) adaptive filtering algorithm, which is specifically designed for high-noise environments such as airflow turbulence, material impact, and mechanical vibration in fly ash sorting industrial sites. The algorithm's filtering window length is synchronized with the data acquisition frequency of 100ms / time. During operation, the algorithm first calculates the moving mean and moving variance of the original acquired data. Data with instantaneous fluctuations exceeding three times the moving mean are identified as abnormal interference data and are removed. At the same time, the effective data is centered, scaled, and normalized to eliminate calculation biases caused by different physical dimensions and different acquisition ranges. Finally, it outputs standard chemical condition data with high signal-to-noise ratio and high consistency.

[0058] The airflow sorting sensor acquisition system is arranged according to the functional areas of the sorting chamber. Flow velocity sensors and pressure sensors are arranged in the rising airflow area, the swirling classification area, and the settling discharge area, respectively. Air volume and torque sensors are set at the exhaust fan outlet and the return air duct. Moisture content, particle size distribution, and fineness online detection sensors are installed at the feed inlet and the discharge outlet, respectively. All sensors are synchronously triggered and acquired using a unified clock source. The acquired data is transmitted in real time to the preprocessing unit via fieldbus. Through hardware filtering and software noise reduction, the effective working condition characteristics are preserved to the greatest extent.

[0059] In this embodiment of the invention, the deviation tracing stage is based on the standardized operation data preprocessed in the parameter acquisition stage. The industrial control system compares the fineness data of the finished fine powder with the standard sieve residue range of primary and secondary fly ash in real time, quantifies and calculates the fineness deviation value and the deviation change rate, and completes the automatic determination of the working condition level in combination with the correlation model.

[0060] The system binds material operating condition fluctuation data and airflow field parameter drift data to distinguish the two types of causes of fineness fluctuations. The two types of causes are material property fluctuations caused by changes in moisture content, particle size, and feed flow rate, and airflow separation field equipment parameter drifts caused by negative pressure offset, flow field unevenness, and classifier wheel speed deviation. This enables the separation of abnormal equipment failures from normal material operating condition fluctuations and pinpoints the root cause of control triggers.

[0061] Fineness deviation quantification calculation uses the absolute value of the difference between the real-time value and the median of the standard interval as the deviation amplitude, and simultaneously calculates the change in deviation amplitude per unit time as the deviation change rate, thereby comprehensively judging the severity of operating condition fluctuations. When judging the operating condition level, the fineness deviation amplitude is first compared with the preset threshold, and then the deviation change rate is combined to distinguish whether the fluctuation is a slow drift or a sudden anomaly, and further distinguish between two causes: material property fluctuations and equipment parameter drifts. If it is determined to be a material property fluctuation, the system only performs negative pressure adaptive compensation; if it is determined to be an equipment parameter drift, the system synchronously triggers equipment status verification to quickly locate the abnormal operation of key components such as the classifier wheel, induced draft fan, and return air duct.

[0062] Formula for quantifying fineness deviation:

[0063]

[0064] It represents the magnitude of fineness deviation, which is the absolute deviation between the real-time fineness of the finished fine powder and the standard fineness, directly reflecting the degree to which the fineness of the finished product deviates from the qualified standard;

[0065] This represents the real-time sieve residue value of the finished fine powder, which is the actual measured value of the sieve residue of the finished fine powder obtained through real-time monitoring during the production process.

[0066] This represents the standard sieve residue benchmark value for fly ash, which is the intermediate standard value of qualified sieve residue for Grade I and Grade II fly ash determined according to national standards.

[0067] It represents the rate of change of fineness deviation, used to measure how quickly the fineness deviation changes over time, and reflects the severity of fluctuations in the sorting conditions;

[0068] It represents the change in the magnitude of the fineness deviation per unit time. It is the increase or decrease in the magnitude of the fineness deviation within a fixed time period, reflecting the magnitude of the deviation change.

[0069] This indicates the calculation time interval, which is a fixed calculation duration preset by the system and synchronized with the data acquisition cycle.

[0070] The distinction between material property fluctuations and equipment parameter drift is made by cross-comparison of multi-dimensional parameters. The system simultaneously retrieves three types of material parameters: feed flow rate, raw ash moisture content, and original particle size distribution, as well as three types of equipment parameters: negative pressure in the sorting chamber, speed of the classifying wheel, and airflow velocity. The parameter data from multiple consecutive collection cycles are compared horizontally and correlated vertically. When material parameters change synchronously and equipment parameters remain constant, a material property fluctuation flag is executed. When material parameters do not change significantly and equipment parameters show a shift, an equipment parameter drift flag is executed.

[0071] In this embodiment of the invention, the negative pressure control stage is based on the working condition level and the cause of fineness fluctuation determined in the deviation tracing stage. The hierarchical incremental PID feedback adjustment algorithm is used to complete the dynamic correction of the sorting negative pressure. The corresponding parameter adjustment logic is matched according to the working condition level to realize the sorting negative pressure calibration and avoid the problems of sorting working condition oscillation, sorting failure and high energy consumption.

[0072] Under slight fluctuation conditions, an incremental correction mode is adopted to control the frequency adjustment of the induced draft fan within 0.5% to maintain the airflow field balance in the sorting chamber. Under moderate deviation conditions, the negative pressure compensation amount is calculated through a layered coupling model, and the incremental adjustment step size is adaptively matched in combination with real-time feed flow rate and raw ash moisture content data to smoothly calibrate the negative pressure threshold.

[0073] Under severe abnormal operating conditions, a gradient progressive incremental adjustment mode is adopted, and the operating condition early warning linkage mechanism is activated simultaneously. The negative pressure parameters are corrected in stages and layers to stabilize the state of the sorting airflow field. The algorithm incremental step size can be adaptively and dynamically adjusted according to the operating condition deviation level and the real-time state of the airflow field, so as to realize the fine dynamic control of the sorting negative pressure and improve energy utilization efficiency.

[0074] The graded incremental PID feedback regulation algorithm is an adaptive intelligent control algorithm that deeply integrates the characteristics of fly ash sorting process. The algorithm input port receives three core feature data in real time: fineness deviation amplitude, deviation change rate, and operating condition level. The kernel constructs three independently optimized subsets of PID parameters to adapt to slight fluctuation operating conditions, moderate deviation operating conditions, and severe abnormal operating conditions, respectively. The parameter set for slight fluctuation operating conditions is Kp=0.75, Ki=0.18, Kd=0.12; the parameter set for moderate deviation operating conditions is Kp=0.9, Ki=0.22, Kd=0.08; and the parameter set for severe abnormal operating conditions is Kp=1.05, Ki=0.25, Kd=0.05.

[0075] The algorithm automatically switches parameter combinations according to the operating condition level, and adjusts the step size by combining real-time negative pressure deviation and airflow uniformity nonlinear calculation, and outputs two sets of control commands: the frequency adjustment of the induced draft fan and the negative pressure compensation value of the sorting chamber.

[0076] Graded incremental PID negative pressure regulation formula:

[0077]

[0078] The increment of the negative pressure adjustment for the kth time represents the specific adjustment value of the negative pressure in the sorting chamber or the frequency of the induced draft fan output by the system.

[0079] This represents the proportionality coefficient, which is set with specific parameters according to different working conditions. Its main function is to quickly respond to real-time changes in fineness deviation.

[0080] This represents the integral coefficient, which is set to match different operating conditions and offset levels. It is used to eliminate static errors caused by fineness deviations and ensure adjustment accuracy.

[0081] It represents the differential coefficient, which is adapted to various operating condition fluctuation settings and can suppress airflow field fluctuations caused by deviation changes in advance;

[0082] This represents the fineness deviation value of the kth sampling, which is the deviation between the fineness of the finished product and the standard value within the current sampling period.

[0083] This represents the fineness deviation value of the (k-1)th sampling, which is the deviation value between the fineness of the finished product and the standard value in the previous sampling period;

[0084] This represents the fineness deviation value of the (k-2)th sampling, which is the deviation between the fineness of the finished product and the standard value within the first two sampling periods.

[0085] Once the aforementioned working condition early warning linkage mechanism is activated, the system will push audible and visual early warning signals and details of abnormal working conditions to the industrial control terminal in real time, and simultaneously record all parameter data of the current abnormal working condition to form a traceability file. At the same time, it will automatically reduce the feed flow rate to the safe operating range. During the entire process of negative pressure gradient adjustment, the system will simultaneously verify the stability of the airflow field after each level of adjustment. If the abnormal state is not improved after three consecutive adjustments, the system will automatically trigger the protective flow stabilization program, temporarily lock the core equipment parameters, and prompt on-site personnel to check the equipment status.

[0086] In this embodiment of the invention, the flow field adaptation stage is based on the dynamic correction result of the negative pressure of the sorting completed in the negative pressure control stage. Through the principles of wind-driven airflow sorting and centrifugal classification, a three-zone coordinated control system of rising airflow zone, swirling classification zone, and settling discharge zone is established to coordinately adapt and control the flow field parameters of the three functional areas of the sorting chamber.

[0087] For the rising airflow zone, the opening of the air inlet distribution plate is dynamically adjusted to ensure that the rising airflow uniformly lifts the fly ash particles; for the swirling classification zone, the auxiliary speed of the classification wheel is calibrated synchronously to stabilize the centrifugal separation airflow field and lock the critical particle size for particle classification; for the settling discharge zone, the air volume ratio of the return air duct is adaptively adjusted to balance the settling airflow pressure and avoid the mixing of coarse and fine particles and uneven particle stratification.

[0088] The three-zone coordinated control system adopts a time-series linkage control logic. First, it adjusts the opening of the uniform distribution plate in the rising airflow zone to ensure that the airflow lifting force is uniform and stable. Then, it synchronously adjusts the speed of the classifying wheel in the swirling classification zone to lock the critical particle size of the classification in the target range. Finally, it adaptively matches the return air volume ratio of the settling discharge zone according to the operating status of the first two zones. The interval between the three-zone adjustment actions does not exceed 100ms.

[0089] During the control process, the airflow velocity difference and pressure fluctuation value of each region are monitored in real time. When the pressure difference between regions exceeds the set threshold, the fine-tuning and flow stabilization operation is automatically triggered to maintain the uniformity and stability of the flow field in the sorting chamber.

[0090] The critical particle size for particle grading is locked based on the target fineness requirement of the finished fly ash. It is combined with the real-time calculation and matching of the current negative pressure value of the sorting chamber and the intensity of the swirling airflow. After the negative pressure and flow field parameters are corrected, the system will gradually fine-tune the speed of the grading wheel to make the critical particle size stably match the grading standard corresponding to the 45μm sieve residue. During the flow field adaptation process, the critical particle size will be dynamically calibrated according to the real-time feedback of the finished product fineness to always maintain a high degree of matching between the grading threshold and the target fineness.

[0091] In this embodiment of the invention, the closed-loop verification stage is based on the flow field collaborative adaptation and control operation completed in the flow field adaptation stage. A three-level control mechanism of adjustment effect verification, system error compensation, and secondary correction is established. The finished product fineness, coarse powder fineness content, and airflow uniformity data of the subsequent three operation cycles are continuously collected to evaluate the parameter adjustment effect.

[0092] A single operation cycle is the complete process cycle of fly ash from feeding to completion of sorting and discharge. The duration of a single cycle is uniformly set according to the actual processing capacity of the production line. Multiple sets of data are collected at a fixed frequency within each cycle, and the average value is taken as the valid data for that cycle. The continuous collection of three operation cycles can completely cover the stable flow field stage and the stable finished product discharge stage after parameter adjustment, avoiding misjudgment of the adjustment effect due to single sampling deviation.

[0093] When the fineness of the finished product returns to the standard range, the current operating parameters and flow field ratio parameters are locked to maintain steady-state operation. When there is a small residual deviation in the fineness of the finished product, the system calculates the adjustment error coefficient and performs pre-compensation on the subsequent PID incremental adjustment parameters to offset the inherent adjustment lag error of the equipment. When the adjustment effect does not meet the standard requirements, the system automatically backtracks the material operating conditions, airflow field parameters, and negative pressure operation data to re-determine the cause of the deviation, and starts secondary graded adjustment to eliminate the accumulation of parameter adjustment error.

[0094] The three-level control mechanism is equipped with clear quantitative judgment criteria. The control effect is verified by the condition that the fineness of the finished product falls within the standard range for three consecutive working cycles and the coarse powder content is lower than the limit value. The system error compensation is for small residual deviations. The average deviation is calculated as the error coefficient and the coefficient is directly added to the output of the next PID control to achieve pre-compensation.

[0095] The secondary correction is initiated when the fineness still fails to meet the standard after two consecutive adjustments. At this time, the system comprehensively reviews the operating condition data, negative pressure data, and flow field data of the past 10 cycles, re-examines the source of deviation and determines the operating condition, and then performs graded adjustment again using the corrected control parameters.

[0096] After the secondary graded adjustment is started, the system first clears the temporary calculation data of the previous round of adjustment, then retrieves the full-dimensional working condition dataset within the specified time period, and re-completes the working condition level determination according to the initial deviation tracing process. After the determination is completed, the negative pressure correction strategy of the corresponding gear is called, the negative pressure compensation amount and adjustment step size are recalculated, and the negative pressure calibration, flow field adaptation and data verification operations are executed in sequence.

[0097] In this embodiment of the invention, the iterative learning stage is based on the adjustment data, error correction data and working condition record data of the closed-loop verification stage to build a dynamic self-learning database. The dynamic self-learning database has the functions of working condition classification iteration, long-term slow variable compensation and automatic parameter calibration, so as to realize the long-term adaptive operation of the sorting system. The system classifies and stores the optimal negative pressure parameters, three-zone flow field ratio parameters, fineness compliance data and deviation traceability records under different material working conditions and different airflow field states, and constructs a working condition parameter database.

[0098] The dynamic self-learning database performs a multi-level filtering process for data entry. First, it filters datasets whose operating conditions last for a minimum duration, removing parameter records corresponding to short-term abrupt changes in operating conditions. Second, it verifies the continuity of parameters within the dataset, removing records with missing data or abnormal jumps. Finally, it categorizes the filtered datasets according to material operating condition type, airflow field state level, and equipment operating mode. Data in the same category are stored sequentially according to the collection time. The format and fields of the data entering the database are uniformly organized according to the database's preset specifications, and all storage operations are performed according to established classification and sorting rules.

[0099] By combining long-term operating variables such as seasonal humidity changes in the production line, raw coal quality fluctuations, and long-term equipment wear, the hierarchical negative pressure and fineness correlation model is optimized through big data iteration. The benchmark negative pressure threshold, PID increment step size, and three-zone flow field ratio corresponding to each operating condition are updated simultaneously. The system can adaptively match short-term instantaneous operating condition fluctuations with long-term seasonal and equipment aging operating condition changes, thereby improving the intelligent operation level of the production line and reducing manual operation and maintenance costs.

[0100] The dynamic self-learning database adopts a distributed streaming storage and online incremental learning architecture, and integrates four major functional modules: working condition classification storage, long-term slow variable compensation, optimal parameter calibration, and model iterative optimization. It can fully store the optimal negative pressure parameters, three-zone flow field ratio, fineness compliance data, and deviation traceability records of different material properties, airflow field status, equipment operating conditions, and the entire process.

[0101] The database uses a streaming big data real-time processing engine, which receives 24-hour continuous operating data from the production line every day, completes an incremental model update every week, and performs a full parameter confidence interval calibration every month. It can automatically identify and compensate for slow variable interferences such as seasonal humidity changes, long-term drift of raw coal quality, and wear of equipment parts. After iterative optimization, it synchronously updates the optimal range of values ​​for the benchmark negative pressure threshold, PID adjustment step size, and flow field ratio parameters corresponding to each operating condition.

[0102] In addition to a fixed-cycle start, the iterative update of the layered negative pressure and fineness correlation model is also set with adaptive triggering conditions. When the production line experiences continuous fluctuations in operating conditions that exceed the original model's adaptation range, the fineness of the finished product continuously deviates from the standard range, or the core components of the equipment are maintained or replaced, the system will automatically start the model iteration in advance. During the iteration process, newly generated high-quality operating condition data will be fused with historical data for training, and the model's feature weights and parameter mapping relationships will be updated to ensure that the model always adapts to the real-time status of the production line.

[0103] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0104] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for adjusting the fineness of fly ash sorting based on negative pressure feedback, characterized in that, The specific steps include the following: In the working condition modeling stage, a hierarchical multi-working condition coupled correlation model is constructed, a sorting benchmark database is established, the correlation between core parameters is fitted, the working condition judgment interval for fineness deviation is divided, and a dynamic correction mechanism for particle size critical value is set. During the parameter acquisition phase, four core data categories are collected simultaneously: sorting equipment operation, material conditions, airflow field, and finished fine powder quality. After preprocessing to remove abnormal data and standardization, the data is uploaded to the industrial control system. In the deviation tracing stage, the deviation amplitude and rate of change of the fineness of the finished product are quantitatively calculated, and the sorting condition level is determined by combining the correlation model, the causes of deviation are broken down and the root cause of abnormality is located. During the negative pressure control stage, a graded incremental PID feedback control algorithm is adopted, and a negative pressure correction strategy is set for different sorting conditions to adaptively adjust the control step size. During the flow field adaptation stage, a three-region collaborative control system for the sorting cavity is established. The airflow parameters are adjusted according to the airflow characteristics of each functional region to control the critical particle size for particle classification. During the closed-loop verification phase, a closed-loop verification and error correction mechanism is constructed to continuously monitor the quality parameters of the finished fine powder, assess the adjustment effect, and perform parameter locking, error compensation, or secondary adjustment accordingly. During the iterative learning phase, a dynamic self-learning database is built, and the working condition correlation model and control parameters are iteratively optimized based on big data algorithms, with real-time updates of the baseline parameters and control parameters corresponding to each working condition.

2. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 1, characterized in that, The hierarchical multi-condition coupling correlation model has the functions of condition-level threshold calibration, particle size critical value coupling correction, and sorting purity weight compensation. When modeling, the core working condition range of fly ash wind-driven negative pressure sorting is locked, and three types of fluctuating working conditions are selected: gradient feed flow rate, raw ash moisture content, and original particle size distribution. Sorting parameters including airflow field wind speed and swirling airflow intensity are matched, and steady-state operating benchmark data including sorting chamber reference negative pressure, graded airflow velocity, 45μm sieve residue fineness, coarse powder fineness content, and fine powder entrainment rate are calibrated. A three-dimensional nonlinear correlation model of negative pressure threshold, fineness qualification rate, and sorting purity is fitted.

3. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 2, characterized in that, Based on calibration data, a three-dimensional nonlinear correlation model is fitted to the negative pressure threshold, fineness qualification rate, and sorting purity corresponding to light, medium, and heavy working conditions. A stratified judgment standard is formulated, and a standardized fineness deviation threshold range is divided and bound to the corresponding airflow field working conditions. The correlation model has a built-in dynamic correction coefficient for particle size critical value, which can complete the adaptive compensation of negative pressure threshold according to the real-time change of the original fly ash particle size.

4. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 3, characterized in that, The parameter acquisition phase is based on the correlation model and working condition judgment criteria set in the working condition modeling phase to set the data acquisition frequency and build an airflow sorting sensor acquisition system. The core data collected synchronously includes the airflow velocity in multiple areas of the sorting chamber, the return air pressure difference, the airflow torque of the classifying wheel, the real-time negative pressure value of the sorting chamber, the air volume of the induced draft fan, the rotation speed of the classifying wheel, the instantaneous flow rate of the feed, the moisture content of the raw ash, the particle size distribution of the feed, the real-time sieve residue value of the finished fine powder, and the fineness content of the coarse powder.

5. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 4, characterized in that, An adaptive filtering and noise reduction algorithm is used to preprocess the collected data, removing abnormal interference data caused by instantaneous material impact, airflow turbulence disturbance, and equipment vibration during the wind sorting process. The preprocessed valid data is then standardized and uploaded to the industrial control system.

6. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 5, characterized in that, The deviation tracing stage is based on the standardized operation data preprocessed in the parameter acquisition stage. The industrial control system compares the fineness data of the finished fine powder with the standard sieve residue range of primary and secondary fly ash in real time, quantifies the fineness deviation value and the rate of change, and automatically determines the working condition level by combining the correlation model. It binds the material working condition and airflow field parameter data, distinguishes the two types of fineness fluctuation causes: material property fluctuation and equipment parameter drift, and separates equipment failure and abnormal material working condition fluctuations.

7. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 6, characterized in that, The negative pressure control stage is based on the working condition level and the cause of fineness fluctuation determined in the deviation tracing stage, and matches the corresponding adjustment logic to complete the dynamic correction of the sorting negative pressure; under slight fluctuation working conditions, the incremental correction mode is adopted to control the fine adjustment range of the induced draft fan frequency. Under moderate offset conditions, the negative pressure compensation is calculated through the hierarchical multi-condition coupling correlation model. The system combines real-time feed flow rate and raw ash moisture content to adaptively match the incremental adjustment step size. Under severe abnormal operating conditions, the negative pressure parameter is adjusted incrementally in a gradient manner, and the operating condition early warning linkage mechanism is activated simultaneously.

8. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 7, characterized in that, The flow field adaptation stage establishes a three-region coordinated control system based on the negative pressure correction results of the negative pressure control stage: the rising airflow zone, the swirling classification zone, and the settling discharge zone. The opening of the air inlet distribution plate is dynamically adjusted for the rising airflow zone, the rotation speed of the classification wheel is synchronously calibrated for the swirling classification zone, and the air volume ratio of the return air duct is adaptively adjusted for the settling discharge zone.

9. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 8, characterized in that, The closed-loop verification stage establishes a three-level control mechanism of adjustment effect verification, system error compensation, and secondary correction. It continuously collects data on finished product fineness, coarse powder fineness content, and airflow uniformity to evaluate the adjustment effect. When the finished product fineness returns to the standard range, the current operating parameters are locked. If there is a small residual deviation, the adjustment error coefficient is calculated and the subsequent PID incremental adjustment parameters are pre-compensated. If the standard is not met, the operating conditions and equipment parameters are traced back to re-determine the cause of the deviation and a secondary graded adjustment is initiated.

10. The method for adjusting the fineness of fly ash sorting based on negative pressure feedback according to claim 9, characterized in that, The dynamic self-learning database has functions such as working condition classification and iteration, long-term slow variable compensation, and automatic parameter calibration. It classifies and stores the optimal negative pressure parameters, flow field ratio parameters, fineness compliance data, and deviation traceability records under different working conditions. Combined with variables including seasonal humidity changes in the production line, raw coal quality fluctuations, and equipment wear, the hierarchical multi-working-condition coupled correlation model is optimized through big data iteration, and the benchmark negative pressure threshold, PID incremental step size, and flow field ratio parameters corresponding to each working condition are updated synchronously.

Images