Composite dust removal method and system, storage medium and program product

Abstract

The invention provides a composite dust removal method and system, a storage medium and a program product, and relates to the technical field of glass bottle production. By recording the instantaneous minimum value of the pressure difference after ash removal and determining the dynamic baseline value based on the time sequence change trend, the performance aging of the filter bag caused by long-term use can be tracked, so that the ash removal threshold value can be dynamically adjusted. On the basis, through superposing a scrambling variable on a speed raising instruction and iteratively optimizing, an optimal operation strategy can be explored and converged autonomously. Finally, the obtained optimal scrambling quantity is solidified into a subsequent speed increasing instruction, so that the whole dust removal method does not depend on static and one-time parameter setting any more, self-iteration is continuously carried out while filter bag aging and working condition fluctuation are coped, finally, on the premise that the dust removal effect is guaranteed, the effective dust removal interval is prolonged, and the dust removal efficiency is improved. And the operation efficiency and economic benefits of the system are ensured.

Description

Technical Field

[0001] This application relates to the field of glass bottle manufacturing technology, and in particular to a composite dust removal method, system, storage medium, and program product. Background Technology

[0002] In modern industrial production, especially in areas involving the handling of glass waste such as glass bottle manufacturing and recycling, large amounts of dispersed dust are generated during processes such as material crushing, screening, and conveying. This dust not only pollutes the production environment and endangers the occupational health of workers, but may also wear down precision equipment and, under certain conditions, even pose a risk of dust explosion.

[0003] To address these needs, a common approach in related technologies is to employ a control mode that combines step-wise power regulation linked to the production line status with a fixed differential pressure upper limit. Specifically, the dust collection system's fan power is preset to several fixed levels, such as "standby" and "high-power operation." When the associated main production line equipment starts, the control system switches the dust collection fan from the low-power standby level to the preset high-power operation level. During operation, an independent dust removal control unit continuously monitors the filter bag clogging status via a differential pressure sensor. Once the real-time differential pressure value exceeds a constant upper limit threshold set based on equipment safety and design experience, the dust removal system automatically triggers to clean the filter bags. This solution achieves basic on-demand operation and passive protection against filter bag clogging.

[0004] However, as the manufacturing industry becomes increasingly demanding in its requirements for lean management and energy efficiency, the aforementioned control mode, which relies on static, preset parameter sets, reveals its profound limitations under complex and ever-changing actual operating conditions. On one hand, as the core filtration element, the filter bag's substrate undergoes irreversible aging over long-term use, manifesting as a continuous and slow increase in the initial filtration resistance baseline. A fixed dust removal pressure differential threshold cannot respond to this long-term change, resulting in the actual effective dust holding space (i.e., the difference between the upper limit threshold and the real-time baseline value) being continuously compressed in the later stages of filter bag use, leading to increasingly shorter effective dust removal cycles. On the other hand, the sudden and drastic fluctuations in production load (such as the concentrated crushing of a batch of high-density materials) intertwine with this long-term performance degradation trend, making the system's operating state extremely complex. The preset fixed high-power operating level is a product of trade-offs among various typical operating conditions, and it is almost impossible to be the optimal solution under any specific, atypical operating condition. The direct consequence of this strategy's mismatch with operating conditions is that, throughout the entire equipment lifecycle and highly dynamic production tasks, the effective cleaning interval time (i.e., the runtime from the completion of one cleaning cycle to the next cleaning cycle) that the system's unit power consumption can obtain always hovers at a relatively low level that cannot be improved on its own.

[0005] In summary, there are issues such as performance aging and instantaneous dynamic fluctuations in dust-generating conditions, which lead to a decrease in overall operating efficiency. Summary of the Invention

[0006] This application provides a composite dust removal method, system, storage medium, and program product, which extends the effective dust removal interval while ensuring dust removal effect, thereby ensuring the system's operating efficiency and economic benefits.

[0007] In a first aspect, this application provides a composite dust removal method, comprising: acquiring the dust-generating points of a glass waste system; calculating the rate of change of current over time based on the current of the glass bottle production equipment corresponding to the dust-generating points, and obtaining current change rate data; generating an acceleration command when the current change rate data exceeds a preset acceleration threshold, and driving the dust removal equipment to increase its power according to the acceleration command; recording the instantaneous minimum pressure difference value of the inlet and outlet pressure difference after the dust removal equipment completes cleaning; storing multiple instantaneous minimum pressure difference values ​​in a time series to obtain a time series, and determining the dynamic baseline value of the instantaneous minimum pressure difference value changing with time based on the changing trend of the time series; adding the dynamic baseline value to a preset constant pressure difference increment to obtain a dynamic cleaning threshold; repeatedly executing the following steps until a preset number of times; adding a preset disturbance variable to the acceleration command during the next acceleration command execution; recording the time elapsed when the pressure difference reaches the dynamic cleaning threshold; iteratively changing the magnitude or direction of the preset disturbance variable; acquiring the optimal disturbance variable that maximizes the time length after the loop ends; and adding the optimal disturbance variable to subsequent acceleration commands.

[0008] By employing the aforementioned technical solution, the power increase of the dust removal equipment is triggered by calculating the rate of change of current in the glass bottle production equipment. This achieves a rapid and proactive response to dust generation events, avoiding the lag in related differential pressure control and ensuring that dust is effectively captured before it diffuses. By recording the instantaneous minimum value of the differential pressure after cleaning and determining the dynamic baseline value based on its time-series change trend, the performance aging of the filter bags due to long-term use can be effectively tracked, allowing the cleaning threshold to be dynamically adjusted, thereby maintaining a constant effective dust holding space. Based on this, by superimposing perturbation variables on the acceleration command and iteratively optimizing, using the "time elapsed until the differential pressure reaches the dynamic cleaning threshold" as a performance indicator, the system can autonomously explore and converge to an optimal operating strategy. Finally, the obtained optimal perturbation variables are solidified into subsequent acceleration commands, so that the entire dust removal method no longer relies on static, one-time parameter settings, but continuously iterates itself while coping with filter bag aging and operating condition fluctuations. Ultimately, while ensuring dust removal efficiency, the effective cleaning interval is extended, guaranteeing the system's operating efficiency and economic benefits.

[0009] In conjunction with some embodiments of the first aspect, in some embodiments, the step of generating an acceleration command specifically includes: determining the operating state of the corresponding glass bottle production equipment based on the current; when the operating state is stable, performing time-frequency transformation on the current to obtain the current time spectrum; extracting frequency domain feature vectors from the current time spectrum, and combining the frequency domain feature vectors with the current change rate data to obtain composite feature data; and generating an acceleration command when the composite feature data meets the preset acceleration trigger conditions.

[0010] By adopting the above technical solution and ensuring the equipment operates in a stable state, misjudgments caused by the massive surge current generated during equipment start-up and shutdown are effectively mitigated, avoiding ineffective speed-up and energy waste. Subsequently, under stable operating conditions, the current undergoes time-frequency transformation, and frequency domain feature vectors are extracted from the obtained current time-frequency spectrum. Finally, this in-depth frequency domain feature vector is combined with the original current change rate data to form composite feature data. This combination ensures that the speed-up triggering decision is no longer based on an isolated, easily disturbed indicator, but rather on a comprehensive profile of the equipment's true load state, thereby reducing the probability of erroneous triggering.

[0011] In conjunction with some embodiments of the first aspect, in some embodiments, the step of generating an acceleration command specifically includes: a preset command strategy library, which stores multiple predefined command templates, and each command template is associated with a specific composite feature data pattern; when the composite feature data meets the preset acceleration triggering conditions, the composite feature data is used as a query index; a search is performed in the command strategy library to retrieve a specific command template that matches the current pattern of the composite feature data; and the specific command template is instantiated as an acceleration command.

[0012] By adopting the above technical solution, and by pre-setting an instruction strategy library that stores multiple instruction templates, and associating each template with a specific composite feature data pattern, the system can transform abstract operating condition identification results into specific, executable control strategies. When a specific composite feature data meets the triggering condition, it no longer triggers a general action, but rather acts as a query index to precisely retrieve the corresponding pre-optimized specific instruction template from the instruction strategy library. This ensures that every power adjustment is highly targeted, thereby achieving refined management of energy consumption while meeting dust removal requirements.

[0013] In conjunction with some embodiments of the first aspect, in some embodiments, the step of superimposing a preset perturbation variable onto the acceleration command specifically includes: superimposing the preset perturbation variable onto the corresponding dedicated perturbation variable in the acceleration command; after the loop ends, the step of superimposing the optimal perturbation variable onto the subsequent acceleration command specifically includes: after the loop ends, superimposing the optimal perturbation variable onto the corresponding dedicated perturbation variable in the subsequent acceleration command.

[0014] By adopting the above technical solution, and by clearly defining the superposition target of the perturbation variable as a "dedicated perturbation variable" associated with a specific command template, the optimization process is no longer about blindly adjusting the final acceleration command, but rather about fine-tuning the intrinsic parameters of the response strategy for a specific operating condition (i.e., the command template). The system can perform optimization iterations for each independent command template and solidify its optimization results (optimal perturbation variable) onto the perturbation variable dedicated to that template.

[0015] In conjunction with some embodiments of the first aspect, in some embodiments, the step of performing time-frequency transformation on the current to obtain the current time spectrum specifically includes: processing the acquired current time series signal through a short-time Fourier transform algorithm to generate a current time spectrum; determining the frequency range of interest for the current time spectrum, which is pre-calibrated based on the characteristic mechanical vibration frequency generated when the glass waste is broken; and retaining only the spectral energy distribution of the current time spectrum within the frequency range of interest to obtain the current time spectrum.

[0016] By employing the aforementioned technical solution, a "frequency range of interest" is defined for the time-frequency spectrum. This range is pre-calibrated based on the characteristic mechanical vibration frequencies generated by the core dust-generating behavior (fracture of glass waste). By retaining and outputting only the spectral energy distribution within this frequency range of interest, all frequency band information irrelevant to the target event is effectively eliminated. This allows the final generated current time-frequency spectrum to more directly and clearly reflect the intensity and pattern of the actual dust-generating process. Consequently, the characterization capability and stability of the frequency domain feature vector subsequently extracted from this high-quality time-frequency spectrum are improved.

[0017] In conjunction with some embodiments of the first aspect, in some embodiments, after recording the time elapsed when the pressure difference reaches the dynamic dust removal threshold, the method further includes: integrating the current change rate data within the time length to obtain the total periodic dust generation load corresponding to the time length; after the cycle ends, obtaining the optimal perturbation variable that maximizes the time length, specifically including: after the cycle ends, using the obtained total periodic dust generation load, dividing the time length by the corresponding total periodic dust generation load to obtain the unit load running time; selecting the preset perturbation variable corresponding to the maximum value of the unit load running time as the optimal perturbation variable.

[0018] By adopting the above technical solution, in the optimization loop, by integrating the current change rate data, a total periodic dust generation load corresponding to the duration of each test can be obtained. Based on this, by dividing the duration by the corresponding total dust generation load, the performance index of unit load runtime is obtained. The essence of this index is "the effective runtime that the system can maintain when processing a unit amount of dust." Selecting the maximum value of this index to determine the optimal perturbation variable avoids misjudgments caused by accidentally obtaining a longer duration under low dust generation loads, ensuring the fairness and robustness of the optimization process.

[0019] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of superimposing a preset perturbation variable into the acceleration command during the next acceleration command execution, the method further includes: pre-recording and saving a baseline performance snapshot; real-time monitoring of one or more safety critical indicators; if any safety critical indicator exceeds a preset safety threshold, immediately terminating the current cycle; and rolling back according to the baseline performance snapshot.

[0020] By adopting the above technical solution, the system has a clear rollback point by pre-recording and saving a baseline performance snapshot, providing reversibility assurance for all exploratory operations. During optimization testing, real-time monitoring of safety-critical indicators can immediately halt any dangerous attempts that might lead to equipment damage or malfunction, avoiding catastrophic consequences. After the optimization cycle ends, the obtained optimal performance is compared with the baseline performance snapshot. This effectively prevents the optimization process from getting stuck in local optima or being interfered with by abnormal data, thus avoiding arriving at an optimal solution that is actually worse than its intended performance. If a significant performance degradation occurs, the system will automatically discard the result and roll back to the initial state.

[0021] In a second aspect, this application provides an electronic device comprising: one or more processors and a memory; the memory being coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, and the one or more processors invoking the computer instructions to cause the electronic device to perform the method described in the first aspect and any possible implementation thereof.

[0022] Thirdly, this application provides a computer program product containing instructions that, when run on an electronic device, cause the electronic device to perform the method described in the first aspect and any possible implementation thereof.

[0023] Fourthly, this application provides a computer-readable storage medium including instructions that, when executed on an electronic device, cause the electronic device to perform the method described in the first aspect and any possible implementation thereof.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By calculating the rate of change of current in the glass bottle production equipment to trigger a power increase in the dust removal equipment, a rapid and proactive response to dust generation events is achieved, avoiding the lag in related differential pressure control and ensuring that dust is effectively captured before it diffuses. By recording the instantaneous minimum value of the differential pressure after cleaning and determining the dynamic baseline value based on its time-series change trend, the performance aging of filter bags due to long-term use can be effectively tracked, allowing for dynamic adjustment of the cleaning threshold and maintaining a constant effective dust holding space. Based on this, by superimposing perturbation variables onto the acceleration command and iteratively optimizing, using the "time elapsed until the differential pressure reaches the dynamic cleaning threshold" as a performance indicator, the system can autonomously explore and converge to an optimal operating strategy. Finally, the obtained optimal perturbation variables are solidified into subsequent acceleration commands, so that the entire dust removal method no longer relies on static, one-time parameter settings, but continuously iterates itself while coping with filter bag aging and operating condition fluctuations. Ultimately, while ensuring dust removal efficiency, the effective cleaning interval is extended, guaranteeing the system's operating efficiency and economic benefits.

[0025] 2. By ensuring the equipment operates in a stable state, misjudgments caused by the massive surge current generated during equipment start-up and shutdown are effectively mitigated, preventing ineffective speed increases and energy waste. Subsequently, under stable operating conditions, a time-frequency transformation is performed on the current, extracting frequency domain feature vectors from the resulting current time-frequency spectrum. Finally, this in-depth frequency domain feature vector is combined with the original current change rate data to form composite feature data. This combination ensures that the speed-up triggering decision is no longer based on an isolated, easily disturbed indicator, but rather on a comprehensive profile of the equipment's true load state, thereby reducing the probability of erroneous triggering.

[0026] 3. By pre-setting an instruction strategy library that stores multiple instruction templates and associating each template with a specific composite feature data pattern, the system can transform abstract operating condition identification results into concrete, executable control strategies. When a specific composite feature data meets the triggering condition, it no longer triggers a general action, but rather acts as a query index to precisely retrieve the corresponding pre-optimized specific instruction template from the instruction strategy library. This ensures that every power adjustment is highly targeted, thereby achieving refined management of energy consumption while meeting dust removal requirements. Attached Figure Description

[0027] Figure 1 This is a schematic flowchart of a composite dust removal method in an embodiment of this application; Figure 2 This is another schematic diagram of the composite dust removal method in the embodiments of this application; Figure 3 This is an exemplary hardware structure diagram of a composite dust removal system in an embodiment of this application. Detailed Implementation

[0028] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0030] Please see Figure 1 , Figure 1 This is a schematic flowchart of a composite dust removal method in an embodiment of this application; S101. Obtain the dust generation points of the glass waste system; The “glass waste system” refers to a complete production line for processing waste glass bottles and other materials, which typically includes equipment for feeding, crushing, screening, and conveying. The “dust-generating point” refers to the specific physical location or equipment in the system that generates a large amount of dust due to material processing, such as the crusher cavity, the conveyor belt discharge port, and the vibrating screen.

[0031] Specifically, this step is typically performed during the system initialization or configuration phase. Technicians will clearly identify the core dust-generating equipment in the control system based on the process flow. For example, in a system containing crusher A, conveyor belt B, and vibrating screen C, the feed inlets of A, B, and C are the dust-generating points that are "acquired" and recorded. The purpose of this step is to shift subsequent monitoring and control from a general overview of the entire workshop to a precise focus on the source of dust generation, laying the physical foundation for subsequent predictive control based on the status of the source equipment.

[0032] S102. Based on the current of the glass bottle production equipment corresponding to the dust generation point, calculate the rate of change of current over time and obtain the current change rate data. Among them, "glass bottle production equipment" refers to power equipment that directly processes materials at the dust-generating point, usually driven by an electric motor, such as a crusher motor; "current" refers to the armature current consumed by the equipment during operation; "rate of change of current over time" refers to the rate at which the current value changes with time, i.e., dI / dt, which is used to characterize the instantaneous change of the equipment load.

[0033] Specifically, the control system uses current transformers installed on the power supply lines of dust-generating equipment (such as crusher motors) to collect current signals in real time at a high sampling frequency (e.g., 1000 times per second). The algorithm within the controller continuously performs first-order differential or more complex differential operations on the collected current values ​​to obtain a real-time current change rate data stream that reflects the drastic changes in load. When a large number of glass bottles are fed into the crusher, the motor load increases dramatically, causing a sharp rise in current. At this time, the calculated current change rate value will show a significant peak.

[0034] S103. When the current change rate data exceeds the preset speed-up threshold, a speed-up command is generated, and the dust removal equipment is driven to increase its power according to the speed-up command. Among them, the "speed-up command" is a control signal generated by the main controller and sent to the dust removal equipment drive unit (such as the frequency converter); "dust removal equipment" mainly refers to the core power unit of the dust removal system. Specifically, the current change rate data stream is compared with a preset speed-up threshold. Once the currently calculated current change rate value exceeds this threshold, the controller determines that a large-scale dust generation event is about to occur. At this time, it immediately generates a speed-up command, which may contain information such as target frequency, target power, or a preset power curve. This command is sent to the frequency converter controlling the fan, which then increases the output frequency, driving the fan motor to speed up, thereby increasing the extraction power and negative pressure of the entire dust removal system in a short period of time to cope with the approaching dust cloud.

[0035] S104. After the dust removal equipment completes the dust removal process, record the instantaneous minimum pressure difference between the inlet and outlet after the cycle ends. Among them, "inlet and outlet pressure difference" refers to the pressure difference between the dirty air chamber and the clean air chamber of the dust collector, which is a key indicator for measuring the degree of filter bag blockage; "instantaneous minimum pressure difference value" refers to the pressure difference reading at the moment when a dust cleaning procedure has just been completed and the filter bag has returned to the cleanest state in the current cycle.

[0036] Specifically, once the system detects that the differential pressure has reached the cleaning threshold and initiates the cleaning procedure, the controller waits for the cleaning action to complete (e.g., the backflushing valve closes). At the moment of completion, the controller immediately reads and records the differential pressure sensor value. This value represents the inherent resistance of the filter bag at its optimal permeability state after this cleaning, reflecting the filter bag's "health condition" or "degree of aging" at that current moment.

[0037] S105. Store multiple instantaneous minimum differential pressure values ​​in a time series to obtain a time series, and determine the dynamic baseline value of the instantaneous minimum differential pressure value as time changes based on the changing trend of the time series. Here, "time series" refers to a data queue formed by arranging multiple instantaneous minimum differential pressure values ​​in chronological order of their occurrence; "trend" refers to the overall direction of the data queue over a long period of time, which is usually a slow increase; and "dynamic baseline value" refers to a value that can smoothly reflect this long-term trend of change after mathematical fitting or filtering of the time series.

[0038] Specifically, once enough data points are collected (e.g., data from the past 100 cleaning cycles), the system applies algorithms such as moving averages, exponential smoothing, or linear regression to analyze the time series. The purpose of the algorithm is to filter out random noise caused by fluctuations in the effectiveness of individual cleaning cycles and extract the core trend of the differential pressure baseline value slowly increasing over time (e.g., several weeks or months). This smoothly changing trend value is the dynamic baseline value.

[0039] S106. Add the dynamic baseline value to the preset constant differential pressure increment to obtain the dynamic dust removal threshold. Among them, "preset constant differential pressure increment" is a physical quantity representing the "effective dust holding space" of the filter bag, which indicates how much additional dust is allowed to accumulate above the baseline; "dynamic cleaning threshold" refers to the real-time changing upper limit of differential pressure used to trigger the next cleaning, which is obtained by adding the dynamic baseline value and this constant increment.

[0040] It should be noted that, under a fixed threshold, as the baseline rises, the effective dust-holding space is continuously compressed, resulting in shorter and shorter dust removal cycles. By allowing the threshold to rise along with the baseline, it is ensured that the amount of dust that can be handled in each dust removal cycle (i.e., the pressure difference increment) remains at a constant and economical level, thereby stabilizing the production cycle and optimizing dust removal energy consumption.

[0041] S107. Repeat the following steps until the preset number of times; The following steps, S108 to S109, are an iterative optimization process.

[0042] S108. When the speed-up command is executed again, a preset disturbance variable is added to the speed-up command; and the time elapsed when the pressure difference reaches the dynamic dust removal threshold is recorded. Specifically, in each iteration of the optimization loop, the system selects a value (e.g., -10%) from a preset set of disturbance variables. When the acceleration event in S103 is triggered, the system adds this -10% power disturbance to the original acceleration command, making the peak power of the wind turbine 10% lower than the normal value. Simultaneously, the system starts a timer to precisely record the time taken from the end of the last dust removal to the pressure difference reaching the dynamic dust removal threshold calculated in S106 again. This "time length" is the core indicator for evaluating the effectiveness of this disturbance strategy.

[0043] S109. Iteratively change the magnitude or direction of the preset disturbance variable; Specifically, after completing a test using a -10% perturbation variable, the system enters the next iteration. At this point, it automatically selects the next perturbation variable, for example, -5%. Then, the entire process of S108 is repeated: a -5% power perturbation is added during acceleration, and the new "time length" is recorded. This process continues, and the system will sequentially test all preset perturbation variables (such as -10%, -5%, 0, +5%, +10%, etc.), or intelligently select the next test point according to a certain optimization algorithm (such as hill climbing), until the preset number of iterations is reached.

[0044] S110. After the loop ends, obtain the optimal perturbation variable that maximizes the time length. Specifically, the optimization process ends. The controller reviews and compares all the "time length" values ​​recorded in each loop. For example, it might find that when the disturbance variable is +5%, the recorded time length is 35 minutes, the longest of all tests. Therefore, the system determines that "+5%" is the "optimal disturbance variable" found in this optimization loop. This result means that, under the current operating conditions and filter bag status, moderately increasing the acceleration power by 5% can achieve the best operating efficiency.

[0045] It should be noted that theoretical models often deviate from actual operating conditions, and experience can vary from person to person. This solution is based entirely on data obtained through actual testing under real operating conditions. The parameter that yields the longest effective runtime is considered the optimal parameter.

[0046] S111: Add the optimal perturbation variable to the subsequent acceleration command.

[0047] Among them, "subsequent speed-up instructions" refers to all speed-up instructions generated by S103 after the current optimization cycle ends and the system resumes normal production operation.

[0048] Specifically, after determining the optimal perturbation variable in S110, the system stores this value as a resident, global adjustment parameter. From then on, in normal, non-optimal operating mode, for each acceleration event triggered by the rate of change of current, the system will automatically and permanently add this power adjustment while generating the basic acceleration command. This means that the system has updated its operating strategy to a better state through self-learning and has solidified this learning result into daily operation.

[0049] It is evident that triggering the power increase of the dust removal equipment by calculating the rate of change of current in the glass bottle production equipment achieves a rapid and proactive response to dust generation events, avoiding the lag in related differential pressure control and ensuring that dust is effectively captured before it diffuses. By recording the instantaneous minimum value of the differential pressure after cleaning and determining the dynamic baseline value based on its time series change trend, the performance aging of the filter bags due to long-term use can be effectively tracked, allowing the cleaning threshold to be dynamically adjusted, thereby maintaining a constant effective dust holding space. On this basis, by superimposing perturbation variables on the acceleration command and iteratively optimizing, using the "time elapsed when the differential pressure reaches the dynamic cleaning threshold" as a performance indicator, the system can autonomously explore and converge to an optimal operating strategy. Finally, the obtained optimal perturbation variables are solidified into subsequent acceleration commands, so that the entire dust removal method no longer relies on static, one-time parameter settings, but continuously iterates itself while dealing with filter bag aging and operating condition fluctuations, ultimately extending the effective cleaning interval while ensuring dust removal effect, thus guaranteeing the system's operating efficiency and economic benefits.

[0050] The above technical solution, while ensuring dust removal efficiency, extends the effective dust removal interval, guaranteeing system operating efficiency and economic benefits. However, in actual use, the principle in step S102—that the rate of change of current dI / dt can elegantly predict material input and dust generation—is flawed. An increase in dI / dt equals the start of material processing, which means dust is about to increase. The core of the two-phase glass waste system—the crusher—is a heavy-duty induction motor. Its surge current at startup can reach 5-8 times the rated current, generating a dI / dt that is an unparalleled peak value compared to any normal material input. If the system does not differentiate between these, each equipment start-up and shutdown will trigger a meaningless, maximum-power "predictive acceleration," making predictive control extremely unstable. The fan may react sensitively to equipment start-ups and shutdowns, and even occasional large pieces of material, not only failing to save energy but also increasing energy consumption and equipment wear. This turns "prediction" into "disturbance" based on erroneous signals.

[0051] Please see Figure 2 , Figure 2 This is another schematic diagram of the composite dust removal method in the embodiments of this application; Therefore, in some embodiments, the generation of the acceleration command in step S103 may optionally be implemented as follows: S201. Determine the operating status of the corresponding glass bottle production equipment based on the current. Among them, "operating status" refers to the discrete labels obtained after classifying the current operating conditions of the production equipment, such as "stopped", "starting", "stable operation", "no load" or "fault".

[0052] Specifically, the control system determines the equipment status by analyzing the macroscopic characteristics of the real-time acquired current signals. For example, when the current value fluctuates slightly near zero, the status is determined to be "stopped"; when the current experiences a huge instantaneous peak value far exceeding the rated value and then quickly drops back to a higher level, the status is determined to be "starting up"; when the current value fluctuates stably near a higher average value, the status is determined to be "stable operation". Only in the "stable operation" state will subsequent refined analysis be triggered, thus effectively avoiding the interference of drastic current fluctuations caused by non-production states (such as start-up and shutdown) on dust generation prediction.

[0053] S202. Under stable operating conditions, perform time-frequency transformation on the current to obtain the current time spectrum; Among them, "current time spectrum" refers to the two-dimensional data matrix or image generated by the transformation, with time on the horizontal axis, frequency on the vertical axis, and color or brightness representing the energy intensity at that frequency at that moment.

[0054] Specifically, the control system extracts a short segment of the most recent current time series data (e.g., data from the past second) and processes it using algorithms such as Short-Time Fourier Transform (STFT) or Wavelet Transform. The result is a time-frequency spectrum, which acts like a "score," revealing not only the overall "volume" (current amplitude) of the signal but also the constituent "notes" (frequency components) and how these "notes" are played over time. For example, continuously crushed homogeneous material might appear as a smooth, bright line on the time-frequency spectrum with energy concentrated in a specific frequency band.

[0055] In some specific embodiments: time-frequency transformation of the current specifically includes: S2021. The collected current time series signal is processed by the short-time Fourier transform algorithm to generate a current time spectrum. Among them, "current time series signal" refers to a series of current values ​​that are continuously collected and recorded by a current sensor at fixed time intervals; "Short-Time Fourier Transform" (STFT) is a classic signal processing algorithm. Its core idea is to divide a long time series signal into multiple shorter, time-continuous windows, and perform Fourier transform on the signal in each window to obtain the frequency distribution information in each time period.

[0056] Specifically, the control system extracts a fixed-length analysis window (e.g., 1024 sampling points) from the real-time acquired current data stream. It then applies the STFT algorithm to this window of data for calculation. This calculation outputs a two-dimensional data matrix, where one dimension represents time (corresponding to the movement of the window) and the other represents frequency. The values ​​in the matrix represent the signal energy intensity at that specific time and frequency point. This energy matrix, when visualized, is a colorful time-frequency spectrum graph that dynamically shows how the "frequency fingerprint" of the device load evolves over time.

[0057] S2022. Determine the frequency range of interest for the current time spectrum. The frequency range of interest is pre-calibrated based on the characteristic mechanical vibration frequency generated when the glass waste is crushed. Among them, "frequency range of interest" refers to a known frequency range that is of particular interest and contains core information; "characteristic mechanical vibration frequency" refers to the mechanical vibration with a specific frequency range that will inevitably be caused when a glass bottle is impacted, squeezed, and broken in the crusher. This vibration will be transmitted to the motor through the mechanical structure and will eventually be reflected in the motor current in the form of fluctuations at a specific frequency; "pre-calibration" means that this range is not calculated online, but is determined in advance and fixed in the system configuration through experiments and data analysis during the system installation and commissioning phase.

[0058] Specifically, technicians will allow the crusher to process typical glass waste under controlled conditions, while simultaneously monitoring the complete time spectrum of the current signal using a spectrum analyzer or the system's own analysis function. Through multiple experiments, they will discover that despite the presence of 50Hz power frequency interference and other random noise, a significant energy jump always occurs within a relatively fixed frequency band (e.g., 80Hz-120Hz) whenever the glass is effectively broken. This frequency band is then identified as the "frequency range of interest" and written into the controller as a configuration parameter.

[0059] S2023. Only the spectral energy distribution of the current-time spectrum within the frequency range of interest is retained to obtain the current-time spectrum.

[0060] Specifically, this step is a data filtering or "masking" operation. After S2021 generates a complete original time-spectrum matrix covering frequencies from 0 Hz to the Nyquist, the algorithm in this step iterates through this matrix. For each frequency value in the matrix, the algorithm checks whether it falls within the "frequency range of interest" (e.g., 80 Hz - 120 Hz) defined in S2022. If the frequency value is outside this range, all energy values ​​corresponding to that row are either set to zero or discarded. After processing, the output is a "cleaner" time-spectrum with a smaller data volume. Although this new time-spectrum has the same name as the original, its content has been "purified," containing only the core frequency information most relevant to the glass breakage event.

[0061] As can be seen, a "frequency range of interest" was defined for the time-frequency spectrum. This range was pre-calibrated based on the characteristic mechanical vibration frequencies generated by the core dust-generating behavior (fracture of glass waste). By retaining and outputting only the spectral energy distribution within this frequency range of interest, all frequency band information irrelevant to the target event was effectively eliminated. This allows the final generated current time-frequency spectrum to more directly and clearly reflect the intensity and pattern of the actual dust-generating process. Therefore, the characterization capability and stability of the frequency domain feature vector subsequently extracted from this high-quality time-frequency spectrum are improved.

[0062] S203. Extract frequency domain feature vectors from the current time spectrum, and combine the frequency domain feature vectors with the current change rate data to obtain composite feature data; Among them, "frequency domain feature vector" refers to a set of values ​​calculated from the current time spectrum to quantify its key characteristics, such as the total energy of a specific frequency band, spectral centroid, spectral entropy, etc. These values ​​together constitute a multi-dimensional vector; "composite feature data" refers to a new data structure with higher dimensions and more comprehensive information formed by merging the current change rate data representing time domain information and the frequency domain feature vector representing frequency domain information.

[0063] Specifically, after generating the time spectrum, this massive two-dimensional data is not used directly. Instead, it is "dimensionality reduced" and "feature extracted" through algorithms. For example, the algorithm calculates the average energy (feature 1) and the frequency corresponding to the energy peak (feature 2) within the 50-100Hz frequency band of the time spectrum, forming a frequency domain feature vector such as [15.3, 88.7]. At the same time, the system also calculates the current change rate data (from the logic of S102) within the same time period, such as 35.2. Then, the system combines these two to form a composite feature data, such as [15.3, 88.7, 35.2]. This new data point comprehensively describes both the "appearance" (degree of change) and the "intrinsic nature" (vibration mode) of the current event.

[0064] S204. When the composite feature data meets the preset acceleration trigger conditions, an acceleration command is generated.

[0065] Here, the "preset acceleration trigger condition" is no longer a simple numerical threshold, but a complex logical rule or model that applies to multidimensional composite feature data.

[0066] Specifically, the controller compares the composite feature data [15.3, 88.7, 35.2] generated in S203 with preset trigger conditions. These conditions might be a series of logic gates, such as IF(feature 1>10) AND(feature 2>80) AND(feature 3>30) THENTRIGGER; or they might be a pre-trained machine learning model (such as a support vector machine or decision tree) that takes the composite feature data as input and directly outputs a "trigger" or "don't trigger" decision. Only when this multi-dimensional, comprehensive condition is met will the system generate and issue an acceleration command. This decision-making method is far more accurate and reliable than the simple single-threshold comparison in S103.

[0067] As can be seen, by determining the equipment's operating state to be stable, misjudgments caused by the massive surge current generated during equipment start-up and shutdown are effectively shielded, avoiding ineffective speed-up and energy waste. Subsequently, under stable operating conditions, a time-frequency transformation is performed on the current, and a frequency domain feature vector is extracted from the obtained current time-frequency spectrum. Finally, this in-depth frequency domain feature vector is combined with the original current change rate data to form composite feature data. This combination ensures that the speed-up triggering decision is no longer based on an isolated, easily disturbed indicator, but on a comprehensive profile of the equipment's true load state, thereby reducing the probability of erroneous triggering.

[0068] In some specific embodiments, generating the acceleration command in step S204 specifically includes: S2041. Preset instruction strategy library. The instruction strategy library stores multiple predefined instruction templates, and each instruction template is associated with a specific composite feature data pattern. Among them, "instruction strategy library" refers to a pre-configured database or lookup table within the control system, which constitutes the system's "expert knowledge base"; "predefined instruction template" refers to a standardized power adjustment scheme designed for different types of dust generation events. It is not a single target value, but may be structured data containing multiple parameters such as acceleration rate, peak power, holding time, and deceleration rate; "composite feature data pattern" refers to a specific category or "fingerprint" obtained after clustering or classifying the composite feature data generated in Example 2, such as "brief impact mode" and "continuous grinding mode".

[0069] Specifically, engineers will define several typical dust-generating conditions based on historical data and process knowledge. For example: Mode A (bulk material impact): Its composite characteristic data pattern is characterized by an extremely high current change rate and concentrated low-frequency energy. The associated instruction template may be a "fast response template," defined as: increasing power to 95% within 0.5 seconds, maintaining it for 2 seconds, and then rapidly decreasing it.

[0070] Mode B (continuous input of large quantities of fine materials): Its composite characteristic data pattern is characterized by a moderate and continuous rate of change of current and a wide spectral distribution. The associated instruction template may be a "smooth response template," defined as: smoothly increasing the power to 75% within 2 seconds and maintaining it for 15 seconds.

[0071] Once this instruction policy library is established, it provides the system with a direct mapping from "what is the problem" to "what solution to use".

[0072] S2042. When the composite feature data meets the preset acceleration trigger condition, the composite feature data will be used as the query index. Here, "query index" refers to the "keywords" or "key values" used for searching and matching in the instruction strategy library.

[0073] Specifically, once the system determines that the current composite feature data meets the triggering conditions, it does not immediately generate a general acceleration command. Instead, it first performs a "pattern matching" or classification on the current composite feature data (e.g., a multidimensional vector [15.3, 88.7, 35.2]). This process may use a classifier algorithm to determine which "composite feature data pattern" defined in S2041 the vector is closest to. The algorithm's output, i.e., the matched pattern label (e.g., "Pattern A: Large material impact"), is used as a query index for subsequent searches. The core function of this step is to transform real-time, continuous sensor data into a discrete, semantically clear condition label.

[0074] S2043. Search in the instruction strategy library to retrieve a specific instruction template that matches the current pattern of the composite feature data; Among them, "specific instruction template" refers to the predefined power regulation scheme that is uniquely found in the strategy library and corresponds to the current operating mode.

[0075] Specifically, the system uses the query index generated in the previous step (e.g., "Pattern A: Large Material Impact") to search in the instruction strategy library established in S2041. Since each record in the library is a "pattern-template" key-value pair, this search process is very fast and reliable. The system will accurately find the "fast response template" associated with "Pattern A" (increase power to 95%, maintain for 2 seconds, ...) and return it as the search result. At this point, the system has found an optimal response plan pre-set by experts for the currently accurately identified dust generation event.

[0076] S2044. Instantiate a specific instruction template as an acceleration instruction.

[0077] "Instantiation" is the process of transforming an abstract template into concrete, executable instructions. It means filling the parameters defined in the template (such as target power, duration, etc.) into the actual sequence of control commands.

[0078] Specifically, after the controller obtains the "fast response template" retrieved by S2043, it parses the parameters within it. For example, if the template requires "increasing the power to 95% within 0.5 seconds," the controller will generate a series of specific instructions to be sent to the inverter. For instance, over the next 500 milliseconds, the target frequency will be increased by one step every 50 milliseconds until the frequency value corresponding to 95% power is reached. Similarly, it will also set a timer based on the "maintain for 2 seconds" parameter in the template. The final generated command sequence, containing complete timing and target values, and directly executable by the underlying hardware, is the final, "instantiated" speed-up instruction. Compared to the instruction generated by S103, this instruction has richer, more customized time-varying characteristics.

[0079] As can be seen, by pre-setting an instruction strategy library that stores multiple instruction templates and associating each template with a specific composite feature data pattern, the system can transform abstract operating condition identification results into concrete, executable control strategies. When a specific composite feature data meets the triggering condition, it no longer triggers a general action, but rather acts as a query index to precisely retrieve the corresponding pre-optimized specific instruction template from the instruction strategy library. This ensures that every power adjustment is highly targeted, thereby achieving refined management of energy consumption while meeting dust removal requirements.

[0080] In the above embodiments, there is a potential dimensionality mismatch problem. The system generates a structured, also multidimensional, acceleration command from high-dimensional composite feature data; however, the perturbation variable used for optimization is a single-dimensional scalar. This leads to a key technical detail: how a single numerical value is logically and meaningfully applied to a complex command.

[0081] Therefore, based on the above embodiments, the step S108 of superimposing a preset perturbation variable onto the acceleration command can optionally be implemented as follows: S301. Superimpose a preset perturbation variable onto the corresponding dedicated perturbation variable in the acceleration command; Here, "dedicated perturbation variable" refers to a specific adjustment parameter reserved for online optimization within the instruction template of Example 3. For example, the "rapid response template" contains a dedicated perturbation variable _A, and the "smooth response template" contains a dedicated perturbation variable _B, both of which can initially have a value of 0. This parameter is bound to the template and is used to fine-tune the final execution effect of the template.

[0082] Specifically, but within the context of the templated instruction generation method of Implementation Example 3, when a dust-generating event is identified as "Mode A" (large material impact), the system retrieves the corresponding "rapid response template" from the strategy library. At this time, the optimization program provides a current test value, namely the "preset disturbance variable" (e.g., +5%). Operation S301 does not simply add this +5% to the final instruction, but rather adds it to the dedicated disturbance variable _A parameter within the "rapid response template". During instantiation (S2044), the controller reads the template's basic parameters and combines this temporarily adjusted dedicated disturbance variable _A (e.g., 0+5%) to generate the final acceleration instruction for this test. If the next dust-generating event is identified as "Mode B", the same +5% "preset disturbance variable" will be added to the dedicated disturbance variable _B of the "smooth response template" for testing.

[0083] Step S111 can optionally be implemented as S302: After the loop ends, the optimal perturbation variable is superimposed onto the corresponding dedicated perturbation variable in the subsequent acceleration command.

[0084] In this scenario, "optimal perturbation variable" refers to the perturbation variable value that maximizes the "time length" of a specific composite feature data pattern (e.g., "pattern A") after the optimization loop is completed (e.g., optimal perturbation variable _A = +3%). "Specific perturbation variable corresponding to subsequent acceleration instructions" refers to the permanent specific perturbation variable _A parameter stored in the corresponding template (e.g., "fast response template") in the instruction strategy library after the optimization is completed and the system resumes normal operation.

[0085] Specifically, suppose the system, through multiple tests, determines that "+3%" is the "optimal perturbation variable" for "Mode A". At this point, the controller performs a "solidify learning results" operation: it accesses the instruction policy library, finds the "fast response template" associated with "Mode A", and updates the value of its internal permanent parameter, the dedicated perturbation variable _A, from its initial 0 to +3%. This update is permanent. From then on, in normal production operation, whenever the system recognizes "Mode A" and calls the "fast response template", it automatically applies this +3% optimization to the final generated acceleration instruction without further intervention. Similarly, the system can perform another round of optimization for "Mode B" and may find a completely different optimal value (such as -2%), then solidify this value into the dedicated perturbation variable _B of the "smooth response template".

[0086] As can be seen, by clearly defining the superposition target of the perturbation variables as "exclusive perturbation variables" associated with a specific instruction template, the optimization process is no longer blindly adjusting the final acceleration instruction, but rather fine-tuning the intrinsic parameters of the response strategy for a specific operating condition (i.e., the instruction template). The system can perform optimization iterations for each independent instruction template and solidify its optimization results (optimal perturbation variables) onto the perturbation variables exclusive to that template.

[0087] In practical applications, the dimension of using the optimal perturbation variable for the longest time can be affected by fluctuations in production conditions, thus impacting the optimization results. Therefore, in some implementations: In some embodiments, after step S108, the method further includes: S401. Integrate the current change rate data over the time period to obtain the total periodic dust generation load corresponding to the time period. Among them, "integral processing" refers to the mathematical operation of summing up the current change rate data points collected continuously within a certain time period; "total dust load of the cycle" refers to a quantitative value that represents a relative measure or proxy index of the total amount of dust generated by the production equipment within a complete dust removal cycle (i.e., the "time length" recorded in S108).

[0088] Specifically, this step is executed immediately after the "duration" of a test cycle is recorded in S108. The controller retrieves and processes all recorded "current change rate data" within that time period (e.g., 35 minutes from the end of the last dust removal to the trigger of the current dust removal). Since the magnitude of the current change rate is directly related to the severity of material crushing, i.e., the rate of dust generation, it is integrated (i.e., summed) over the entire cycle, and the result is equivalent to the total intensity of "dust generation events" over the entire cycle. For example, a busy cycle with frequent crushing activities will have a high calculated "total dust load" value; conversely, a quiet cycle will have a low total load value.

[0089] Step S110 can optionally be implemented as S402: After the cycle ends, the unit load running time is obtained by dividing the time length by the corresponding total dust generation load using the obtained total dust generation load of the cycle. Among them, "duration of operation per unit load" is a core performance evaluation indicator, which indicates "how long the dust removal system can maintain effective operation under a unit dust generation load", and is essentially an efficiency indicator.

[0090] Specifically, the actual operation is performed in S402, and this calculation is performed after each iteration of the optimization loop (S107). In each test, the system obtains two core data points: the duration (e.g., 35 minutes) and the corresponding total dust generation load per cycle (e.g., 12,000 units). This step divides the former by the latter to obtain the runtime per unit load (35 / 12,000 = 0.0029 minutes / unit load). This value eliminates the influence of fluctuations in production intensity and more accurately reflects the true impact of the "disturbance variable" of the current test on the operating efficiency of the dust removal system.

[0091] S403. Select the preset disturbance variable corresponding to the maximum value of the unit load running time as the optimal disturbance variable.

[0092] The "maximum value of runtime per unit load" refers to the highest value among all the "runtime per unit load" values ​​calculated in all optimization test cycles.

[0093] Specifically, this step is executed after the entire optimization loop (S107) has finished and is the final decision-making stage. The controller compiles a list containing all test results, each item recording (preset perturbation variable, unit load runtime). For example: [(-5%, 0.0025), (0%, 0.0027), (+5%, 0.0031), (+10%, 0.0029)]. The controller iterates through this list to find the largest "unit load runtime," which in this example is 0.0031. Then, the system determines the "preset perturbation variable" corresponding to this maximum value—i.e., "+5%"—as the "optimal perturbation variable" found in this optimization. This result shows that the +5% perturbation setting allows the system to maintain the longest effective operating time when handling the same amount of dust, meaning the system efficiency is the highest.

[0094] As can be seen, in the optimization loop, by integrating the current change rate data, a total periodic dust generation load corresponding to the duration of each test can be obtained. Based on this, by dividing the duration by the corresponding total dust generation load, the performance index of unit load runtime is obtained. This index essentially represents "the effective runtime that the system can maintain when processing a unit amount of dust." Selecting the maximum value of this index to determine the optimal perturbation variable avoids misjudgments caused by accidentally obtaining a longer runtime under low dust generation loads, ensuring the fairness and robustness of the optimization process.

[0095] In some embodiments, after step S118, the method further includes: S501, Pre-record and save a baseline performance snapshot; The "baseline performance snapshot" refers to a complete backup of all key control parameters of the system before the online optimization cycle (S107) is started. This includes not only the "optimal perturbation variable" obtained from the previous optimization, but may also cover a set of core parameters that constitute the system's current "known best" operating state, such as the base power value and response time in the instruction template.

[0096] Specifically, it's equivalent to creating a "system restore point" before conducting a potentially risky experiment. The controller reads all relevant parameters that are currently running stably and are considered safe and performing well, and packages them into a non-volatile memory area. For example, if the current system is running stably with an "optimal perturbation variable" of +2%, then this "+2%" and other relevant control settings will be saved as the core content of the snapshot.

[0097] S502. Real-time monitoring of one or more critical safety indicators; Among them, "key safety indicators" refer to the core physical quantities that can directly or indirectly reflect the health status and operational safety boundaries of the dust removal system and related equipment. These indicators typically include, but are not limited to: the absolute pressure inside the dust collector housing (to prevent excessive negative pressure from causing deformation of the housing or pipes), the operating temperature of the main fan motor, the vibration intensity of the motor, and the monitoring value of the outlet dust concentration (to ensure environmental compliance), etc.

[0098] Specifically, when the system begins testing a new "preset perturbation variable" (S108), this monitoring program reads data from various sensors at a very high temporal resolution (e.g., tens of times per second). These data streams are continuously fed into the controller for comparison with preset safety thresholds. This monitoring process is global; it does not concern itself with whether the current perturbation variable extends the cleaning cycle, but only with whether it puts the system on the verge of danger or violation.

[0099] S503. If any critical safety indicator exceeds the preset safety threshold, the current cycle shall be terminated immediately. Here, "safety threshold" refers to the pre-set limit value for each "safety critical indicator" that must not be exceeded. For example, the safety threshold for motor temperature might be set at 90°C, and the negative pressure threshold for the enclosure might be set at -5000Pa. These values ​​are determined by equipment specifications and process safety procedures.

[0100] Specifically, during real-time monitoring, once the controller detects that the instantaneous value of any safety-critical indicator touches or exceeds its corresponding safety threshold, it will immediately trigger an interrupt signal. For example, when testing a large disturbance variable of +20%, the fan motor temperature spikes from 75°C to 90.1°C in a short period. At this point, the safety logic will immediately determine a violation and forcibly terminate the current test cycle. The system will immediately abandon the evaluation of this +20% disturbance variable, will not wait for the pressure difference to reach the dust removal threshold, and will not record its "duration".

[0101] S504, and roll back according to the baseline performance snapshot.

[0102] "Rollback" refers to the action of forcibly restoring the system's control parameters to the "baseline performance snapshot" state saved in S501.

[0103] Specifically, after aborting the dangerous test with the +20% perturbation variable, the controller will not attempt the next perturbation variable but will immediately execute a rollback procedure. It will read a previously saved baseline snapshot from storage (e.g., containing an "optimal perturbation variable" of +2%) and overwrite all currently running control parameters with these "known safe" parameters. This instantly restores the entire dust removal system to the stable and reliable operating state it was in before the online optimization began. After the rollback, the system may issue an alarm to the operator and pause the entire optimization process, awaiting manual inspection and intervention.

[0104] As can be seen, by pre-recording and saving a baseline performance snapshot, the system has a clear rollback point, providing reversibility assurance for all exploratory operations. During optimization testing, real-time monitoring of safety-critical indicators can immediately halt any dangerous attempts that might damage the equipment or cause malfunctions, avoiding catastrophic consequences. After the optimization cycle ends, the obtained optimal performance is compared with the baseline performance snapshot. This effectively prevents the optimization process from getting stuck in local optima or being interfered with by abnormal data, thus avoiding arriving at an optimal solution that is actually worse than its intended performance. If a significant performance degradation occurs, the system will automatically discard the result and roll back to the initial state.

[0105] The following describes an exemplary composite dust removal system 300 provided in an embodiment of this application. Figure 3 This is an exemplary hardware structure diagram of the composite dust removal system 300 provided in the embodiments of this application.

[0106] In some embodiments, the composite dust removal system 300 is a computer device or includes a computer device. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The network interface communicates with other external terminals or servers via a network connection. In some embodiments, the network interface can be a wired network interface; in some embodiments, it can also be a wireless network interface. When the computer program is executed by the processor, it implements the methods described in the embodiments of this application.

[0107] Those skilled in the art will understand that Figure 3The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0108] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0109] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0110] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0111] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A combined dust removal method, characterized by, The method comprises the following steps: acquiring a dust generation point of a glass waste material system; calculating a rate of change of current over time based on the current of the glass bottle production equipment corresponding to the dust generation point, to obtain current rate of change data; generating a speed-up instruction when the current rate of change data exceeds a preset speed-up threshold, and driving the dust removal equipment to increase power according to the speed-up instruction; recording the instantaneous minimum pressure difference value of the inlet and outlet air pressure difference after the cycle is completed after the dust removal equipment completes dust removal; storing a plurality of the instantaneous minimum pressure difference values in time sequence to obtain a time sequence, and determining a dynamic baseline value of the instantaneous minimum pressure difference value over time based on the change trend of the time sequence; adding the dynamic baseline value to a preset constant pressure difference increment to obtain a dynamic dust removal threshold; repeatedly performing the following steps until a preset number of times; when the next speed-up instruction is executed, superimpose a preset perturbation variable on the speed-up instruction; and record the time length taken by the pressure difference to reach the dynamic dust removal threshold; iteratively changing the size or direction of the preset perturbation variable; after the cycle is completed, acquiring an optimal perturbation variable that makes the time length reach the longest; superimposing the optimal perturbation variable on the subsequent speed-up instruction.

2. The method of claim 1, wherein, The step of generating a speed-up instruction specifically comprises: determining the operating state of the corresponding glass bottle production equipment based on the current; performing time-frequency transformation on the current to obtain a current time-frequency spectrum when the operating state is stable operation; extracting a frequency domain feature vector from the current time-frequency spectrum, and combining the frequency domain feature vector with the current rate of change data to obtain composite feature data; generating the speed-up instruction when the composite feature data meets a preset speed-up trigger condition.

3. The method of claim 2, wherein, The step of generating the speed-up instruction specifically comprises: a preset instruction strategy library, the instruction strategy library stores a plurality of predefined instruction templates, and each instruction template is associated with a specific composite feature data pattern; when the composite feature data meets a preset speed-up trigger condition, the composite feature data is used as a query index; searching in the instruction strategy library to retrieve a specific instruction template that matches the current pattern of the composite feature data; instantiating the specific instruction template as the speed-up instruction.

4. The method of claim 3, wherein, The step of superimposing a preset perturbation variable on the speed-up instruction specifically comprises: superimposing the preset perturbation variable on the corresponding exclusive perturbation variable in the speed-up instruction; After the cycle is completed, the step of superimposing the optimal perturbation variable on the subsequent speed-up instruction specifically comprises: After the cycle is completed, superimpose the optimal perturbation variable on the corresponding exclusive perturbation variable in the subsequent speed-up instruction.

5. The method of claim 2, wherein, The step of performing time-frequency transformation on the current to obtain a current time-frequency spectrum specifically comprises: processing the collected current time series signal through a short-time Fourier transform algorithm to generate a current time-frequency spectrum; determining a frequency range of interest for the current time-frequency spectrum, which is pre-calibrated according to the characteristic mechanical vibration frequency generated when the glass waste material is crushed; only retaining the frequency spectrum energy distribution of the current time-frequency spectrum within the frequency range of interest to obtain the current time-frequency spectrum.

6. The method of claim 1, wherein, After the step of recording the differential pressure and the length of time elapsed when the differential pressure reaches the dynamic ash removal threshold, the method further comprises: During the length of time, the current rate of change data is integrated to obtain a total dust production load corresponding to the length of time; After the loop ends, the step of obtaining an optimal disturbance variable that maximizes the length of time comprises: After the loop ends, the total dust production load is used to obtain a length of time for unit load operation by dividing the length of time by the corresponding total dust production load; The maximum value of the length of time for unit load operation is selected to determine the optimal disturbance variable.

7. The method of claim 1, wherein, After the step of superimposing the preset disturbance variable on the speed-up instruction, the method further comprises: A baseline performance snapshot is recorded and saved in advance; One or more safety-critical indicators are monitored in real time; If any of the safety-critical indicators exceeds a preset safety threshold, the current loop is immediately terminated; And the baseline performance snapshot is rolled back.

8. An electronic device, comprising: The electronic device comprises one or more processors and a memory; the memory is coupled to the one or more processors; the memory is used to store computer program code, the computer program code comprises computer instructions, and the one or more processors invoke the computer instructions to enable the electronic device to perform the method of any one of claims 1-7.

9. A computer program product comprising instructions, characterized in that, When the computer program product is running on the electronic device, the electronic device is enabled to perform the method of any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, characterized in that, When the instructions are running on the electronic device, the electronic device is enabled to perform the method of any one of claims 1-7.

Images