Blowing control method for high pull wheel and pressing knife of packaging equipment

Abstract

The invention relates to the technical field of packaging machinery control, in particular to a blowing control method for a high pull wheel and a pressing knife of packaging equipment, and adopts the technical scheme that a dynamic pollution risk assessment model fusing multi-source sensing data and a machine learning adaptive mechanism is constructed; the conventional fixed-time and fixed-point passive mode of blowing cleaning is thoroughly changed; a designed multi-level collaborative blowing execution strategy realizes intelligent matching between the cleaning intensity and the production takt; image recognition, pressure fluctuation analysis and temperature monitoring are deeply fused, and three-dimensional and multi-dimensional perception of the working face state is provided; the efficiency and pertinence of the cleaning action are remarkably improved, the consumption of compressed air is reduced to the maximum extent while the optimal cleaning effect is guaranteed, the long-term reliability of the equipment under high-load continuous operation and the consistency of the sealing quality are guaranteed through a layered response mechanism, and the sealing quality is improved. The multi-parameter fusion evaluation mode overcomes the defect that a single sensor is prone to interference or one-sided in sensing.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of packaging machinery control, in particular to a blowing control method for high-pulling wheels and pressure knives of a packaging device. BACKGROUND

[0002] In the field of automated packaging equipment, efficient and reliable sealing of packaging materials is a key link to ensure production efficiency and product quality. Among them, heat sealing technology is widely used because of its high sealing strength and fast speed, and its core is to apply appropriate heat and pressure to the packaging material through heating elements to achieve fusion.

[0003] Among them, the packaging equipment with high-pulling wheels and pressure knives working together is a common device to realize continuous heat sealing operation. This technology aims to pull the packaging material by the high-pulling wheels, and apply instantaneous high pressure and heat at the preset position by the pressure knives to complete the sealing action. In this process, the cleanliness of the sealing area directly affects the sealing quality.

[0004] In the prior art, the high-pulling wheels and pressure knives are prone to accumulate molten residues generated by the packaging material due to heat or dust in the environment on their surfaces under long-term high-speed operation. If these attachments are not cleaned in time, they will contaminate the subsequent packaging material, resulting in defects such as virtual sealing, leakage sealing, or uneven sealing strength. At the same time, the uneven accumulation of residues may change the flatness and heat conduction characteristics of the contact surface between the pressure knife and the material, further exacerbating the instability of the sealing quality. In the continuous and high-beat production scene, the above sealing failure risk caused by contaminants will significantly increase the equipment downtime frequency and reduce the overall production efficiency.

[0005] Therefore, how to realize the continuous and effective cleaning of the working surface of the high-pulling wheels and pressure knives has become a technical problem to be solved to improve the reliability and sealing quality of the packaging equipment. SUMMARY

[0006] The purpose of the present application is to provide a blowing control method for high-pulling wheels and pressure knives of a packaging device to solve the problem of sealing quality decline and equipment reliability reduction caused by the accumulation of contaminants on the working surface of high-pulling wheels and pressure knives in the prior art.

[0007] To achieve the above purpose, the present application adopts the following technical solutions: A blowing control method for high-pulling wheels and pressure knives of a packaging device, which is realized based on a closed-loop control system. The closed-loop control system includes a main control unit, an image acquisition module, a pressure sensing module, a temperature sensing module, and a multi-path collaborative blowing execution module. The method includes the following steps: S1. Multi-source data synchronous acquisition: The main control unit synchronously acquires the surface attachment coverage rate collected and calculated by the image acquisition module, the pressure fluctuation variance of the pressing knife contact pressure collected and calculated by the pressure sensing module, and the temperature gradient of the pressing knife working surface collected and calculated by the temperature sensing module in a 1-second cycle. S2. Dynamic Pollution Risk Assessment: The main control unit invokes a dynamically evolving pollution risk assessment model. Based on the surface attachment coverage, pressure fluctuation variance, and temperature gradient obtained in step S1, it calculates a pollution risk index between 0 and 100. The construction and operation process of the dynamically evolving pollution risk assessment model includes: the main control unit has an initial weight vector, which assigns preset initial weight coefficients to the surface attachment coverage, pressure fluctuation variance, and temperature gradient respectively; in each calculation cycle, the main control unit normalizes the three collected parameter values ​​to make them fall within the numerical range of 0 to 1; then, the normalized parameter values ​​are multiplied by the corresponding current weight coefficients, the three products are summed, and then multiplied by 100 to obtain the pollution risk index for the current cycle. The dynamic evolution mechanism of the pollution risk index and weight vector is as follows: The main control unit continuously records the historical sealing quality test results, characterized by the percentage deviation of the measured sealing strength value from the standard value, and establishes a sliding time window spanning the most recent 100 production cycles. At each assessment moment, the main control unit analyzes the correlation between the pollution risk index sequence and the sealing strength deviation sequence within the sliding window, specifically calculating the Pearson correlation coefficient between each parameter sequence and the sealing strength deviation sequence. If the absolute value of the correlation coefficient of a certain parameter exceeds the preset significance threshold for five consecutive assessment cycles, the main control unit initiates the adjustment of the weight of that parameter, increasing the weight coefficient of that parameter by a fixed adjustment step size, while proportionally reducing the weight coefficients of the other two parameters to keep the total weight sum at 1. S3. Multi-level cleaning strategy decision: Based on the preset threshold range of the pollution risk index calculated in step S2, the main control unit decides to adopt the corresponding air blowing cleaning strategy. The air blowing cleaning strategy is divided into three levels: Level 1 is periodic basic cleaning, where the main control unit controls all air blowing units in the multi-channel collaborative air blowing execution module to be turned on synchronously at fixed time intervals, with a fixed start duration of 200 milliseconds; Level 2 is risk-triggered enhanced cleaning, where the main control unit starts enhanced cleaning when the pollution risk index exceeds the first threshold but is lower than the second threshold. It first identifies the parameter with the highest current weight coefficient in the weight vector, and then starts the air blowing unit directly related to that parameter for targeted cleaning. The air blowing duration of enhanced cleaning is extended to 500 milliseconds; Level 3 is emergency intervention deep cleaning, where the main control unit sends a pause signal to the motion controller of the packaging equipment to request a production break when the pollution risk index exceeds the second threshold. After obtaining the break window, it controls all air blowing units to perform linked blowing at the highest pressure and for the longest duration of 1 second. S4. Air blowing command execution: Before the pressing knife performs the heat sealing action, the main control unit generates specific air blowing unit start and stop timing control commands based on the decision results of step S3, and drives the multi-channel collaborative air blowing execution module to execute and complete the cleaning of the working surface.

[0008] Furthermore, the multi-channel coordinated air blowing execution module includes three independent air blowing units, which are used to perform directional cleaning on the working surface of the high puller, the upper working surface of the pressure knife, and the lower working surface of the pressure knife, respectively. Each air blowing unit consists of a solenoid valve, a section of pressure-resistant pipeline, and a specially designed gas nozzle. The outlet of the gas nozzle is designed to be a flat fan shape, and its spray angle is fixed at 30 degrees.

[0009] Furthermore, in the second-level risk-triggered enhanced cleaning, if the weighting coefficient of surface deposit coverage is the highest, the main control unit will prioritize enhancing the cleaning of the air blowing units on the working surface of the high puller and the working surface of the pressure knife; if the weighting coefficient of pressure fluctuation variance is the highest, the main control unit will enhance the cleaning intensity of all three sets of air blowing units.

[0010] Furthermore, in the third-level emergency intervention deep cleaning, the main control unit can also control the air blowing unit to reciprocate along the working surface axis to expand the cleaning range.

[0011] Furthermore, the image acquisition module includes an industrial camera and a ring-shaped LED light source, which are installed directly above the pressure tool working area. The industrial camera acquires images at a rate of 50 frames per second. The image processing subroutine built into the main control unit performs grayscale conversion and Gaussian filtering for each frame of image. Then, an adaptive thresholding algorithm is used to binarize the image to separate the working surface area and the attachment area. Finally, the proportion of white pixels in the binarized image to the pixels in the entire working surface region of interest is calculated and output as the quantized value of the surface attachment coverage.

[0012] Furthermore, the pressure sensing module employs an array of pressure sensors embedded inside the pressure cutter body to directly measure the pressure distribution at the moment of sealing; the main control unit collects pressure waveform data within each pressing cycle and calculates the variance of the waveform data as a measure of pressure fluctuation variance.

[0013] Furthermore, the temperature sensing module uses a non-contact infrared temperature probe pointing towards the center of the working surface of the pressure tool; the main control unit calculates the standard deviation of the temperature values ​​of 10 consecutive sampling points as a representation of the temperature gradient.

[0014] Furthermore, the main control unit is an industrial-grade programmable logic controller or an embedded industrial computer, which is connected to the image acquisition module, pressure sensing module, temperature sensing module, multi-channel collaborative blowing execution module, and motion controller of the packaging equipment via industrial fieldbus or Ethernet.

[0015] Furthermore, the normalization process employs a linear mapping method, where each parameter is mapped based on its preset upper and lower limits for its working range.

[0016] Furthermore, the method operates within a complete work cycle, which begins when the high-pulling wheel pulls the packaging material into place and ends before the multi-channel collaborative air blowing execution module completes the air blowing cleaning and the pressure knife performs the heat sealing action.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention fundamentally changes the traditional passive mode of timed and fixed-point air-blowing cleaning by constructing a dynamic pollution risk assessment model that integrates multi-source sensor data and machine learning adaptive mechanisms. The system can quantify the degree of pollution risk on the working surface in real time and automatically identify the main pollution factors affecting sealing quality at different production stages, thereby achieving a fundamental leap from "indiscriminate timing" to "precise on-demand" cleaning strategies. This significantly improves the efficiency and targeting of cleaning operations, minimizing compressed air consumption while ensuring optimal cleaning results.

[0018] 2. The multi-level collaborative air-blowing execution strategy designed in this invention achieves intelligent matching between cleaning intensity and production cycle time. Basic cleaning ensures the stability of routine operation; risk-triggered enhanced cleaning enables precise intervention at the nascent stage of potential quality problems, preventing issues before they arise; emergency deep cleaning provides the system with the ability to handle sudden severe contamination, and through intermittent window execution with collaborative equipment, it avoids abrupt interruptions to production continuity. This layered response mechanism ensures the long-term reliability of the equipment and the consistency of sealing quality under high-load continuous operation.

[0019] 3. This invention deeply integrates image recognition, pressure fluctuation analysis, and temperature monitoring, providing a three-dimensional, multi-dimensional perception of the working surface condition. Surface deposit coverage directly reflects visible contaminants, pressure fluctuation variance can sensitively capture contact anomalies caused by sticky residues, and temperature gradient can indirectly reflect whether heat conduction is hindered by the contaminant layer. This multi-parameter fusion assessment method overcomes the shortcomings of single sensors, such as susceptibility to interference or incomplete perception, making contamination risk assessment more comprehensive, accurate, and robust, providing a solid data foundation for high-quality sealing decisions. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical solution architecture of the air blowing control method for the high pull roller and pressure knife of the packaging equipment proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the dynamically evolving pollution risk assessment model in this invention; Figure 3This is a logical flow diagram of the multi-level collaborative blowing execution strategy in this invention; Figure 4 This is a logical flowchart of the multi-source sensor data acquisition and processing stage in this invention. Figure 5 This is a schematic diagram illustrating the relationship between the pollution risk index and the threshold response of the air cleaning level in this invention. Detailed Implementation

[0021] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments. Example

[0022] This application provides a specific embodiment of an air blowing control method for a high-pressure roller and pressure knife in a packaging device. For example... Figure 1 As shown, the implementation of this method relies on a highly integrated closed-loop control system. The core architecture of this system includes a main control unit as the computation and command center, an image acquisition module responsible for capturing visual information, a pressure sensing module responsible for sensing mechanical information, a temperature sensing module responsible for monitoring thermal information, and a multi-channel collaborative air blowing execution module responsible for executing the final cleaning action. These modules are connected to the main control unit via an industrial fieldbus, forming a complete loop of real-time data acquisition, intelligent analysis and decision-making, and precise execution feedback. This embodiment will provide a detailed technical description of each component of the system and its collaborative workflow.

[0023] The main control unit is the central processing and decision-making core of the entire system. This unit is typically implemented using an industrial-grade programmable logic controller (PLC) or an embedded industrial computer, and its hardware configuration must meet the requirements of high-speed data processing and real-time control. Specifically, the main control unit contains at least one multi-core processor with a clock speed of no less than 1.8 GHz to ensure real-time computation capabilities for complex algorithm models. It has a memory capacity of no less than 4 GB for storing real-time acquired sensor data streams, historical quality records, and dynamically evolving model parameters. The storage unit uses a solid-state drive (SSD) with a capacity of no less than 128 GB for storing the operating system, control program, historical operation logs, and fault data. The main control unit is equipped with abundant digital and analog input / output interfaces, as well as at least two Ethernet communication ports and one fieldbus interface. One Ethernet port is used to connect the image acquisition module, transmitting high-frame-rate image data streams; the other Ethernet port or fieldbus interface is used to connect the acquisition terminals of the pressure and temperature sensing modules, as well as the motion controller of the packaging equipment itself, to realize the interaction of control commands and status signals. The main control unit runs specially developed control software, which adopts a modular design and mainly includes a system initialization module, a multi-threaded data acquisition and synchronization module, a dynamic pollution risk assessment model calculation module, a multi-level cleaning strategy decision-making module, and an execution command generation and communication module. After the system is powered on, the initialization module first loads all hardware drivers and reads the system configuration parameters stored in non-volatile memory, including the calibration coefficients of each sensor, the response delay compensation values ​​of each solenoid valve in the air blowing execution module, and the initial weight vector of the dynamic pollution risk assessment model. After initialization is completed, the system enters the real-time operation loop.

[0024] The image acquisition module is responsible for continuous visual monitoring of the high-tension roller and the working surface of the pressure cutter to quantify the contamination status of surface deposits. For example... Figure 2As shown, the physical components of this module include a high-resolution industrial area array camera and a ring-shaped LED light source coaxially mounted with it. The industrial camera uses a global shutter model with a resolution of at least 2 megapixels and a stable frame rate of 50 frames per second. The camera lens is a low-distortion fixed-focus lens, with the focal length determined based on the installation distance to ensure the field of view completely covers the working arc surface of the high-pulling roller and the upper and lower working surfaces of the pressure knife. The ring-shaped LED light source has a color temperature of 6500 Kelvin, providing uniform and shadowless illumination. Its brightness can be adjusted by the main control unit to adapt to changes in ambient light. The module is mounted on a precisely calibrated fixed bracket on the packaging equipment frame, positioned directly above the pressure knife working area. The camera's optical axis is perpendicular to the plane of the working surface, and the installation height ensures that the field of view covers the entire monitored area. The camera connects to the main control unit via a gigabit Ethernet cable, using a streaming protocol for transmission. The image processing subroutine within the main control unit performs a standardized processing flow for each incoming image frame. The process first converts the color image to grayscale to reduce data volume and focus on brightness information. Next, a Gaussian filter with a kernel size of 5 pixels × 5 pixels and a standard deviation of 1.2 is applied to the grayscale image for smoothing and denoising to suppress image sensor noise and subtle reflection interference. The denoised image then enters the crucial segmentation stage. The system employs an adaptive thresholding algorithm for image binarization. This algorithm does not use a globally fixed threshold but dynamically calculates a local threshold based on the average brightness of the neighborhood of each pixel in the image. Specifically, a block size of 31 pixels × 31 pixels and a constant offset of -5 are used. After binarization, the image is converted into a black and white binary image, where white pixels represent areas identified by the algorithm as contaminants or abnormal attachments, and black pixels represent clean working surfaces. Finally, the program defines a fixed region of interest (ROI) in the image. This ROI is pre-calibrated to precisely correspond to the projection range of the actual working surfaces of the high-tension roller and pressure knife in the image. The program counts the number of all white pixels within this ROI and calculates their percentage of the total number of pixels in the ROI. This percentage value, representing the surface attachment coverage, is output as a floating-point number between 0 and 100 and stored along with a timestamp in the main control unit's real-time database for use by the pollution risk assessment model. The processing latency for a single frame of image is strictly controlled within 15 milliseconds to meet the system's real-time requirements.

[0025] The pressure sensing module directly senses changes in the mechanical state of the interface between the pressure cutter and the packaging material during the sealing process. Its measurements sensitively reflect uneven or fluctuating contact pressure distribution caused by contaminants. The core of this module is a miniature thin-film pressure sensor array. This array consists of multiple independent piezoresistive sensing units arranged in a matrix, with each sensing unit typically having a sensing area of ​​2 mm × 2 mm. The entire sensor array is embedded within the internal structure of the pressure cutter body, positioned directly below the working surface. The sensing surface is directly coupled to the metal of the cutter body via thermally conductive insulating material to ensure accurate sensing of the pressure transmitted from the working surface. The array density must be sufficient to cover the effective sealing width of the pressure cutter's working surface; for example, for a 15 mm wide pressure cutter, at least eight sensing units should be arranged along its axis. The signal lines from all sensing units are converged to an integrated signal conditioning and analog-to-digital conversion unit. This unit amplifies and filters the signal from each channel and performs synchronous analog-to-digital conversion at a sampling rate of at least 2000 times per second. The converted digital pressure waveform data is uploaded to the main control unit in real time via a high-speed serial peripheral interface or Ethernet. During a complete sealing cycle, from the moment the pressure blade first contacts the packaging material until complete separation, the pressure data analysis subroutine of the main control unit captures the pressure waveform during this time period. The program first performs baseline correction on the waveform to eliminate the influence of sensor zero drift. Then, it calculates the variance of the pressure waveform data as a measure of pressure fluctuation variance. The formula for calculating the variance is: Pressure fluctuation variance equals the sum of squares of the differences between the pressure value at each sampling point in the pressure waveform and the average pressure value at all sampling points, divided by the total number of sampling points. This variance quantifies the instability of pressure during the sealing process; a larger variance indicates a more unstable contact state, potentially caused by slippage or localized adhesion due to sticky contaminants. The calculated pressure fluctuation variance is normalized to a preset reference range, such as the square of 0 to 10 MPa, and used as a key parameter in the contamination risk assessment model.

[0026] The temperature sensing module is responsible for non-contact monitoring of the temperature distribution uniformity of the pressure cutter's working surface. Abnormal temperature gradients can indirectly reflect heat conduction obstruction caused by a contamination layer on the working surface. This module uses a high-precision infrared temperature probe. The probe is selected with a response wavelength of 8 to 14 micrometers to adapt to metal surface temperature measurement, covering a temperature range of 50 to 300 degrees Celsius with a resolution of no less than 0.1 degrees Celsius. The probe is mounted on a bracket with adjustable angle and position, and its optical axis is precisely pointed to a fixed measurement point in the center area of ​​the pressure cutter's working surface, ensuring that the probe can stably measure the same area every time the pressure cutter reaches the sealing position. The probe sends real-time temperature values ​​to the main control unit via analog output or digital communication interface. The temperature data processing subroutine of the main control unit reads the temperature value at a frequency of 20 times per second. To characterize temperature fluctuations or gradients, the program maintains a first-in-first-out temperature data buffer of length 10. Each time a new temperature sample value is acquired, it is stored in the buffer, and the oldest value is discarded. The program then calculates the standard deviation of 10 temperature values ​​within the buffer zone, using this as a measure of the temperature gradient. The standard deviation reflects the dispersion of temperature measurements over a short period; a large dispersion indicates temperature instability, potentially due to uneven heat transfer caused by surface contamination. The calculated standard deviation of the temperature gradient is also normalized, transforming it into a dimensionless parameter between 0 and 1, which is then input into the contamination risk assessment model.

[0027] The multi-channel coordinated air blowing execution module is the final action execution mechanism of the system, responsible for translating the decision commands from the main control unit into physical cleaning actions. For example... Figure 1 and Figure 3As shown, this module comprises three air-blowing units that are completely independent in terms of physical structure, air path, and control. The first air-blowing unit is designed for the working surface of the high-tension roller, with its specially designed gas nozzle installed above the roller side, spraying towards the contact arc surface between the roller and the packaging material. The second and third air-blowing units target the upper and lower working surfaces of the pressure knife, respectively, with nozzles installed on both sides of the pressure knife, spraying perpendicularly towards the upper and lower working surfaces. Each air-blowing unit consists of three key components connected in series: a high-speed two-position normally closed solenoid valve with a response time of less than 10 milliseconds for both opening and closing; a 6 mm inner diameter polyurethane pressure-resistant pipeline with a working pressure tolerance range of 0 to 1 MPa; and a core specially designed gas nozzle. The outlet of the gas nozzle is precision-machined to form a flat fan-shaped structure with a fixed fan angle of 30 degrees. This design allows the ejected compressed air flow to be thin and sheet-like, covering a strip area matching the width of the working surface with high airflow pressure. Simultaneously, due to the small diffusion angle, the airflow remains concentrated within the effective working distance, resulting in high cleaning efficiency. All solenoid valves are connected to the main control unit via digital output modules, receiving direct on / off control. The air source comes from the factory's clean compressed air network, and after being stabilized at 0.6 MPa by a pressure reducing valve, it is distributed to three branches through the main air line. Each branch is also equipped with a precision pressure regulating valve and a flow meter for independently adjusting and monitoring the blowing pressure and flow rate of each branch, ensuring consistent cleaning results.

[0028] The dynamically evolving pollution risk assessment model is the intelligent core of the entire system, and its principle framework is as follows: Figure 2 As shown. The model runs within the main control unit and is triggered at fixed intervals of 1 second. At the beginning of each cycle, the model synchronously acquires the latest preprocessed data from three sensing modules: surface deposit coverage (parameter A), pressure fluctuation variance (parameter B), and temperature gradient standard deviation (parameter C). First, the main control unit normalizes these three raw parameters, mapping their values ​​to the range of 0 to 1. Normalization uses a linear mapping method, and each parameter has a preset upper and lower limit for its typical operating range. For example, the range of surface deposit coverage is 0 to 5%, the range of pressure fluctuation variance is 0 to 8 MPa², and the range of temperature gradient standard deviation is 0 to 2 degrees Celsius. For parameter A, its normalized value A0 is... norm It equals A divided by 5. For parameter B, its normalized value B0 norm It equals B divided by 8. For parameter C, its normalized value C0 norm Equals C divided by 2. If the actual value of any parameter exceeds the preset range, its normalized value is clamped to 0 or 1.

[0029] The model maintains a weight vector W, which contains three weight coefficients, corresponding to the three normalized parameters mentioned above. That is, W equals w. A w Bw C At the initial stage of the system, a preset initial weight vector is loaded, for example, 0.5, 0.3, 0.2, and satisfies w A +w B +w C =1. The formula for calculating the pollution risk index within each calculation period is: PRI=(A norm ×w A +B norm ×w B +C norm ×w C )×100 PRI stands for Pollution Risk Index, which is a dimensionless value between 0 and 100. This value directly reflects the overall pollution risk level of the working face; the higher the value, the greater the risk.

[0030] The core feature of this model lies in the dynamic evolution capability of the weight vector W. The evolution mechanism relies on learning from historical sealing quality feedback. After each sealing action, the main control unit obtains the measured sealing strength value from the quality inspection unit of the packaging equipment. The system calculates the percentage deviation of this measured value from a preset standard value, denoted as the quality deviation Q. The system establishes a sliding time window covering the most recent 100 production cycles, continuously storing the contamination risk index PRI sequence and the quality deviation Q sequence corresponding to each cycle. Every 100 cycles, the main control unit initiates a weight evaluation. During the evaluation, the parameter A within the sliding window is calculated. norm Sequence, parameter B norm Sequence, parameter C norm The Pearson correlation coefficient between the sequence and the quality deviation Q sequence is obtained as the correlation coefficient r. A r B r C The Pearson correlation coefficient measures the degree of linear correlation between two variables, with a value between -1 and 1. The system sets a significance threshold for correlation, for example, 0.3. The evaluation logic is: if the absolute value of the correlation coefficient of a certain parameter exceeds 0.3 for five consecutive evaluation periods, then the parameter is considered to have a persistently significant correlation with sealing quality under the current production conditions. The main control unit then initiates an adjustment of the parameter's weight. The adjustment rules are as follows: Assuming the correlation coefficient r of parameter A... A If the adjustment conditions are met, then its weight w is adjusted. AA fixed adjustment step size Δ is added, for example, 0.05. Simultaneously, to maintain a total weight sum of 1, the weights of parameters B and C are reduced proportionally. This mechanism allows the model to automatically allocate more decision weights to the contamination characterization parameters that have been proven by historical data to be most relevant to seal quality degradation, thereby achieving adaptive and personalized model tuning. This enables the model to adapt to changes in contamination characteristics under different packaging materials and environmental conditions.

[0031] Based on the pollution risk index PRI output by the dynamically evolving pollution risk assessment model, the system executes a multi-level coordinated air blowing strategy. The logical flow and threshold response relationship of this strategy are as follows: Figure 3 and Figure 5 As shown, the system presets two key thresholds: a first threshold T1 and a second threshold T2, where T1 is less than T2. ​​For example, T1 is set to 30 and T2 is set to 70. The generation and execution of the air-blowing cleaning command are thus divided into three levels.

[0032] The first level is periodic basic cleaning. This is a safety net strategy, executed at fixed time intervals regardless of the contamination risk index, to maintain a basic level of cleanliness on the work surface and prevent the slow accumulation of contaminants. The main control unit controls three sets of air blowing units, which are activated synchronously at fixed time intervals, such as every 10 packaging cycles. The activation duration is fixed at 200 milliseconds. This instruction is directly triggered by the timer module of the main control unit, without going through the risk assessment model, but the execution record is included in the system log.

[0033] The second level is risk-triggered enhanced cleaning. When the current pollution risk index PRI calculated by the dynamically evolving pollution risk assessment model meets the condition that PRI is greater than or equal to T1 and less than T2, the main control unit immediately activates the cleaning strategy for this level. In this mode, the cleaning actions are targeted. The main control unit first checks the current weight vector W and identifies the key parameter with the highest weight coefficient. Then, it activates the air blowing unit most directly related to that parameter for focused cleaning. The decision logic is specified as follows: if w A If the maximum surface contamination coverage is the most significant risk indicator, the main control unit determines that the contamination is primarily composed of visible solid dust or particles. At this point, a control command is generated to prioritize enhancing the cleaning of the air blowing units targeting the high-pressure roller working surface and the pressure cutter working surface. The command specifies that before the next basic cleaning cycle, these two air blowing units should be activated immediately, with the blowing duration extended to 500 milliseconds and the blowing pressure increased by 10%. If w BIf the maximum value, i.e., the pressure fluctuation variance, is the primary indicator, the main control unit determines that there may be viscous liquid residue or film adhesion that is difficult to see with the naked eye. At this point, a control command is generated, simultaneously increasing the cleaning intensity of all three air-blowing units. The command is as follows: immediately trigger the simultaneous activation of all three air-blowing units, extend the blowing time to 500 milliseconds, and uniformly increase the blowing pressure by 15% to utilize stronger airflow to remove the adhered substances. If w C At its maximum, the air blowing is enhanced on the upper and lower working surfaces of the pressure cutter, with a focus on ensuring the cleanliness of the heat conduction surfaces. This level of cleaning is inserted during continuous production operation, aiming for precise intervention when risks first emerge.

[0034] Level 3 is emergency intervention deep cleaning. When the contamination risk index PRI is greater than or equal to T2, it indicates that the working surface contamination is very severe, which is highly likely to cause the upcoming sealing action to fail. The main control unit immediately takes the highest level of response. First, the main control unit sends a high-priority pause request signal to the motion controller of the packaging equipment through the communication interface, along with an estimated cleaning time, such as 1.5 seconds. After completing the current packaging cycle, the motion controller will enter a short production pause and send an pause window confirmation signal back to the main control unit. Upon receiving the confirmation signal, the main control unit immediately executes the deep cleaning program. This program controls all three sets of air blowing units to simultaneously activate and perform linked blowing at the highest pressure allowed by the system, such as 0.8 MPa, with a blowing duration set to 1 second. To further expand the cleaning range, the main control unit can also drive a miniature linear module mounted on the bracket of the air blowing nozzle through an additional control interface, allowing the air blowing unit to perform a short reciprocating movement, such as 20 mm, in the axial direction of the working surface. This mobile blowing can cover a wider area, ensuring thorough removal of stubborn contaminants. After deep cleaning is completed, the main control unit notifies the motion controller to resume production. This strategy achieves thorough cleaning without abruptly disrupting the production cycle by requesting equipment cooperation to obtain a cleaning time window.

[0035] The control process within a complete work cycle is executed according to a strict time sequence. The cycle begins when the packaging material is brought into place by the high-pulling wheel of the packaging equipment, ready for heat sealing. The first step of the process involves the main control unit synchronously acquiring the latest surface attachment coverage calculated by the image acquisition module, the latest pressure fluctuation variance reported by the pressure sensing module, and the latest temperature gradient standard deviation reported by the temperature sensing module through its data acquisition thread. This synchronization is achieved through hardware interrupts or precise software timestamps, ensuring that the three parameters correspond to the working surface state at the same physical moment. The second step involves the main control unit calling the dynamically evolving pollution risk assessment model subroutine. This subroutine takes into account the three normalized parameter values ​​and the currently stored weight vector, calculates the pollution risk index (PRI) for this cycle according to a predetermined formula, and updates the stored weight vector value. The third step involves the strategy decision module determining the threshold range into which the calculated PRI value falls. The decision module also considers the maximum value information of the current weight vector and, according to the aforementioned logic, decides whether to adopt Level 1 basic cleaning, Level 2 risk-triggered enhanced cleaning, or Level 3 emergency intervention deep cleaning. The decision-making module generates specific control commands, including the number of the air-blowing unit requiring action, the absolute or relative delay of air blowing activation, the duration of air blowing, and the required air blowing pressure. In the fourth step of the process, at a precise time point before the pressure cutter begins to press down and perform the heat-sealing action—typically with a 50-100 millisecond cleaning time window—the main control unit, through its digital output module, converts the control commands into on / off electrical signals for the solenoid valves, precisely driving the multi-channel collaborative air-blowing execution modules. After the air blowing action is completed, the system immediately reports the execution status. Subsequently, the pressure cutter performs the heat sealing, completing the packaging. Simultaneously, the quality inspection results of this sealing are later fed back to the system to update the historical quality database and drive the next evolutionary evaluation of the model weights. The entire process ensures that every sealing action is performed on a work surface that has undergone intelligent evaluation and cleaning, forming a complete closed loop from perception to decision-making to execution and learning.

[0036] This embodiment details the system's hardware configuration, data processing algorithms, intelligent model operation mechanism, and multi-level execution strategy, providing a complete and engineering-feasible technical solution. By deeply integrating multi-dimensional sensor information and adaptive learning algorithms, this system achieves precise, on-demand, and intelligent control over the cleaning of the working surfaces of the core components of the packaging equipment, fundamentally improving the stability of sealing quality and the reliability of equipment operation.

Claims

1. A method for blow control of a high pull wheel and a presser knife of a packaging apparatus, characterized by, The method is realized based on a closed-loop control system, which comprises a master control unit, an image acquisition module, a pressure sensing module, a temperature sensing module and a multi-path collaborative blowing execution module; the method comprises the following steps: S1, multi-source data synchronous acquisition: the master control unit synchronously acquires the surface adherent coverage calculated by the image acquisition module, the pressure fluctuation variance of the contact pressure calculated by the pressure sensing module and the temperature gradient of the working surface of the pressure knife calculated by the temperature sensing module at a period of 1 second; S2, dynamic pollution risk assessment: the master control unit calls a dynamically evolving pollution risk assessment model to calculate a pollution risk index between 0 and 100 based on the surface adherent coverage, the pressure fluctuation variance and the temperature gradient obtained in step S1; the construction and operation process of the dynamically evolving pollution risk assessment model comprises that the master control unit is internally provided with an initial weight vector, which respectively assigns preset initial weight coefficients to the surface adherent coverage, the pressure fluctuation variance and the temperature gradient; in each calculation period, the master control unit normalizes the three parameter values to fall within the value range of 0 to 1; then, the normalized parameter values are respectively multiplied by the corresponding current weight coefficients, and the three product results are summed and multiplied by 100 to obtain the pollution risk index of the current period; the dynamic evolution mechanism of the weight vector is that the master control unit continuously records the historical sealing quality detection results represented by the deviation percentage of the measured value of the sealing strength relative to the standard value, and establishes a sliding time window with a time span of the last 100 production periods; at each evaluation time, the master control unit analyzes the correlation between the pollution risk index sequence and the sealing strength deviation sequence in the sliding window, and specifically calculates the Pearson correlation coefficient of each parameter sequence and the sealing strength deviation sequence; if the absolute value of the correlation coefficient of a certain parameter exceeds a preset significance threshold for 5 consecutive evaluation periods, the master control unit starts to adjust the weight of the parameter, increases the weight coefficient of the parameter by a fixed adjustment step, and proportionally reduces the weight coefficients of the other two parameters to maintain the total weight sum as 1. S3, multi-level cleaning strategy decision: the host unit decides to adopt the corresponding blowing cleaning strategy according to the preset threshold interval of the pollution risk index calculated in step S2; the blowing cleaning strategy is divided into three levels: the first level is periodic basic cleaning, the host unit controls all blowing units in the multi-path collaborative blowing execution module to be opened synchronously at a fixed time interval, and the opening time is fixed at 200 milliseconds; the second level is risk-triggered enhanced cleaning, when the pollution risk index exceeds the first threshold but is lower than the second threshold, the host unit starts enhanced cleaning, first identifies the parameter with the highest current weight coefficient in the weight vector, and then starts the blowing unit directly related to the parameter for targeted cleaning, and the blowing time of enhanced cleaning is extended to 500 milliseconds; the third level is emergency intervention deep cleaning, when the pollution risk index exceeds the second threshold, the host unit sends a pause signal to the motion controller of the packaging equipment to request production pause, and after obtaining the pause window, controls all blowing units to perform joint blowing with the highest pressure and the longest time of 1 second. S4, blowing instruction execution: before the hot sealing action of the pressure knife, the host unit generates specific blowing unit start-stop timing control instructions according to the decision result of step S3, and drives the multi-path collaborative blowing execution module to execute, completing the work surface cleaning.

2. The air blowing control method for the high pull wheel and the presser of the packing equipment according to claim 1, characterized in that: The multi-path collaborative blowing execution module includes three groups of independent blowing units, which are used for directional cleaning of the high-lifting wheel work surface, the pressure knife upper work surface and the pressure knife lower work surface respectively; each group of blowing units is composed of an electromagnetic valve, a pressure-resistant pipeline and a specially designed gas nozzle; the outlet of the gas nozzle is designed as a flat fan shape, and the jet angle is fixed at 30 degrees.

3. The air blowing control method for the high pull wheel and the presser of the packing equipment according to claim 2, characterized in that: In the second level risk-triggered enhanced cleaning, if the weight coefficient of the surface attachment coverage is the highest, the host unit preferentially enhances the cleaning of the blowing units of the high-lifting wheel work surface and the pressure knife upper work surface; If the weight coefficient of the pressure fluctuation variance is the highest, the host unit enhances the cleaning intensity of all three groups of blowing units.

4. The air blowing control method for the high pull wheel and the presser knife of the packing apparatus according to claim 1, characterized in that: In the third level emergency intervention deep cleaning, the host unit can also control the blowing units to move back and forth along the work surface axis to expand the cleaning range.

5. The air control method for the high pull wheel and the presser knife of the packing equipment according to claim 1, characterized in that: The image acquisition module includes an industrial camera and a ring LED light source, which are installed directly above the pressure knife working area; the industrial camera acquires images at a rate of 50 frames per second; the image processing subroutine built in the host unit performs grayscale and Gaussian filter denoising on each frame of image, then uses an adaptive threshold algorithm for image binarization to separate the work surface area and the attachment area, and finally calculates the proportion of white pixel points in the binarized image to the entire work surface region of interest pixel points as the quantitative value of the surface attachment coverage.

6. The air control method for the high pull wheel and the presser knife of the packing equipment according to claim 1, characterized in that: The pressure sensing module uses an embedded pressure sensor array installed inside the pressure knife body to directly measure the pressure distribution at the sealing moment; the host unit acquires pressure waveform data during each pressure cycle, and calculates the variance of the waveform data as a measure of pressure fluctuation variance.

7. The air control method for the high pull wheel and the presser knife of the packing equipment according to claim 1, characterized in that: The temperature sensing module adopts a non-contact infrared temperature measuring probe pointing to the center area of the working surface of the pressure knife.

8. The air control method for the high pull wheel and the presser knife of the packing equipment according to claim 1, characterized in that: The main control unit is an industrial programmable logic controller or an embedded industrial computer, which is connected with the image acquisition module, the pressure sensing module, the temperature sensing module, the multi-path collaborative blowing execution module and the motion controller of the packaging equipment through an industrial field bus or Ethernet.

9. The air control method for the high pull wheel and the presser knife of the packing apparatus according to claim 1, characterized in that: The normalization processing adopts a linear mapping method, and each parameter is mapped and calculated according to its preset upper limit value and lower limit value of the working range.

10. The air control method for the high pull wheel and the presser knife of the packing equipment according to claim 1, characterized in that: The method runs in a complete working cycle, which starts from the high pull wheel dragging the packaging material to the position and ends before the multi-path collaborative blowing execution module completes the blowing cleaning and the pressure knife performs the heat sealing action.

Images