An intelligent control system for automatic filling and capping

Abstract

The present application relates to the technical field of automation control, and in particular to an intelligent control system for automatic filling and capping, which solves the technical problem of easy misjudgment of abnormal identification in the prior art. The system comprises: a resistance value determination unit for determining the single-bottle equivalent cap screwing resistance value of a servo capping main shaft for each packaging bottle; each packaging bottle is filled by a corresponding filling pump before capping, a disturbance decoupling processing unit for performing disturbance decoupling processing according to the single-bottle equivalent cap screwing resistance value of each filling pump, generating a consumable global resistance vector and a single-channel friction resistance vector; the consumable global resistance vector is used to represent the overall resistance change trend of each filling pump under different time sliding batches, and the single-channel friction resistance vector is used to represent the friction resistance difference corresponding to different filling pumps within the same time sliding batch; an adaptive compensation control unit for adaptive compensation control according to the consumable global resistance vector and the single-channel friction resistance vector.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and more specifically to an intelligent control system for automatic filling and capping. Background Technology

[0002] In the automated production of small-dose liquid formulations such as oral liquids, production lines are typically equipped with multi-head filling pumps, star wheel conveyor turntables, and servo capping mechanisms. These mechanisms complete the sealing of the packaging bottles through continuous filling and capping actions. Due to inherent height tolerances in the manufacturing process of glass bottles, these tolerances can cause a shift in the triggering timing of the servo spindle contacting the bottle opening. Traditional control methods that rely solely on fixed positions are insufficient to accurately identify whether the bottle is in position. Furthermore, manual sampling and broken bottle rejection frequently occur during production line operation. If the underlying shift register within the controller continues to advance according to a fixed mechanical rhythm, the data queue recorded by the controller will misalign with the actual position of the bottles on the conveyor belt, leading to the failure of subsequent data-based source traceability.

[0003] Furthermore, during continuous operation of small-dose medication filling pumps, wear on the seals can cause minute, thread-like leaks when the pump stops dispensing. These leaked medications, carried by the conveyor turntable, adhere to the external threads of specific batches of bottles, acting as additional lubricant. Downstream capping stations experience a sudden drop in localized frictional resistance when handling these bottles with leaked threads. This sudden decrease in friction means that, within the normally set torque limit, the servo spindle encounters no expected resistance during downward pressure, leading to over-rotation and abnormal compression, ultimately crushing the bottle neck. Simultaneously, the diurnal temperature drop in the workshop causes the entire batch of plastic packaging materials to harden, resulting in a baseline increase and global drift in overall capping friction. In actual measurements, the increase in global environmental resistance due to temperature and the sudden drop in localized friction caused by single-channel leaks—two opposing factors—multiply and superimpose.

[0004] Existing single-machine control systems cannot effectively separate these two types of synchronously occurring product interference data. When a fixed absolute difference or variance is used to set the alarm threshold, the overall amplification of the global baseline will inevitably lead to the synchronous amplification of the absolute value of the decrease in resistance of the local liquid-contaminated channel. This makes traditional early warning identification prone to misjudgment, leading to misjudged shutdowns or continuous bottle breakage accidents. Summary of the Invention

[0005] To address the technical problem of easy misjudgment in anomaly identification in existing technologies, the present invention aims to provide an intelligent control system for automatic filling and capping, the specific technical solution of which is as follows: This invention provides an intelligent control system for automatic filling and capping, comprising: The resistance value determination unit is used to determine the equivalent capping resistance value of a single bottle for each packaging bottle by the servo capping spindle; each packaging bottle is filled by its corresponding filling pump before capping, and the equivalent capping resistance value of a single bottle is used to characterize the average resistance of the servo capping spindle in a single capping process. The interference decoupling processing unit is used to perform interference decoupling processing based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, and generate a global resistance vector and a single-channel friction resistance vector for consumables. The global resistance vector of consumables is used to characterize the overall resistance change trend of each filling pump under different time sliding batches, and the single-channel friction resistance vector is used to characterize the friction resistance difference of different filling pumps within the same time sliding batch. An adaptive compensation control unit is used to perform adaptive compensation control based on the global resistance vector of the consumable and the single-channel frictional resistance vector.

[0006] Optionally, the resistance value determination unit includes: The entity state determination module is used to obtain the peak value of the downward current of the servo capping spindle in each single capping process, compare the peak value of the downward current with the preset idling current threshold, and determine the entity state of the capping station in each single capping process; the entity state is used to characterize whether there is a packaging bottle at the capping station in the corresponding single capping process. The resistance value calculation module is used to obtain the rotational torque, rotational angle, and total downward displacement of the servo capping spindle in the single capping process when the physical state characterizes the presence of a packaged bottle, and to determine the equivalent capping resistance value of a single bottle based on the rotational torque, the rotational angle, and the total downward displacement. The traceability writing module is used to trace the equivalent capping resistance value of a single bottle to the filling pump number based on the conveying trajectory station register, determine the corresponding filling pump number identifier, and write the equivalent capping resistance value of a single bottle to the channel buffer queue corresponding to the filling pump number identifier.

[0007] Optionally, the entity state determination module is specifically used for: During the no-load trial operation of the equipment, the maximum torque current value of multiple no-load pressure actions is extracted to determine the preset no-load current threshold. If the peak value of the down-pressure current is less than the preset idling current threshold, it is determined that the physical state characterization does not contain a packaging bottle, and a packaging bottle missing pulse signal is generated. In response to the packaging bottle missing pulse signal, a null value overwrite operation is performed on the corresponding data bit in the conveying trajectory station register. If the peak value of the down-current is greater than or equal to the preset idling current threshold, it is determined that the physical state characterizes the presence of a packaging bottle.

[0008] Optionally, the resistance value calculation module is specifically used for: When the physical state characterizes the presence of a packaged bottle, the rotational torque, rotational angle, and total downward displacement of the servo capping spindle are obtained during the single capping process. When the total downward displacement is greater than the preset effective displacement threshold, the equivalent capping resistance of a single bottle is calculated based on the rotational torque, the rotational angle, and the total downward displacement to determine the equivalent capping resistance value of a single bottle. When the total downward displacement is less than or equal to the preset effective displacement threshold, a packaging bottle missing pulse signal is generated.

[0009] Optionally, the source tracing writing module is specifically used for: Extract the data index number of the packaging bottle corresponding to the single bottle equivalent capping resistance value from the conveying trajectory station register, and perform a reverse search in the historical data sequence of the conveying trajectory station register according to the data index number to locate the filling station record that performs the liquid injection process on the packaging bottle; Extract the filling pump number identifier bound to the filling station record, and write the single-bottle equivalent capping resistance value of the packaging bottle into the channel buffer queue corresponding to the filling pump number identifier.

[0010] Optionally, the interference decoupling processing unit includes: The historical matrix construction module is used to update historical resistance data based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, and generate a historical capping resistance matrix; the historical capping resistance matrix is ​​used to record the resistance characteristic data of each filling pump channel within multiple consecutive time sliding batches. The rank reduction feature decoupling module is used to perform rank reduction feature decoupling on the historical capping resistance matrix to generate the global resistance vector of the consumable and the single-channel friction resistance vector.

[0011] Optionally, the historical matrix construction module is specifically used for: Whenever the cumulative number of capping actions reaches the preset sliding window size, the historical resistance data of the corresponding time sliding batch is updated. Within the time-sliding batch, the channel buffer queue corresponding to each filling pump is traversed to obtain the equivalent capping resistance value of each bottle in the time-sliding batch for each filling pump channel; The representative value of the single-channel resistance of each filling pump channel in the time-sliding batch is determined based on the equivalent capping resistance value of all single bottles in each filling pump channel in the time-sliding batch. Historical resistance data is updated based on the single-channel resistance representative value of each filling pump channel in the time-sliding batch, generating a historical capping resistance matrix.

[0012] Optionally, the rank reduction feature decoupling module is specifically used for: Perform non-negativity pre-constraint processing on the historical screw cap resistance matrix, assign a preset bias value to the elements in the historical screw cap resistance matrix that do not satisfy the non-negativity condition constraint, and generate a non-negativity constraint matrix. The non-negative constraint matrix is ​​asynchronously factored using a rank-reducing feature decoupling algorithm to generate the global resistance vector of the consumables and the single-channel friction resistance vector.

[0013] Optionally, the adaptive compensation control unit includes: The scaling and index generation module is used to perform benchmark scaling on the global resistance vector of the consumables and the single-channel frictional resistance vector respectively, to generate a normalized consumables resistance vector and a benchmark-scaled channel resistance vector, and to determine the consumables resistance time difference based on the normalized consumables resistance vector, and to determine the channel frictional resistance range based on the benchmark-scaled channel resistance vector; the consumables resistance time difference is used to characterize the overall resistance offset of the whole machine consumables under continuous time sliding batch; the channel frictional resistance range is used to characterize the degree of dispersion of frictional resistance corresponding to different filling pumps within the same time sliding batch. The compensation control execution module is used to perform adaptive compensation control based on the time difference of the consumable resistance and the range of the channel friction resistance.

[0014] Optionally, the compensation control execution module is specifically used for: When the channel friction resistance range is less than a preset dispersion threshold and the consumable resistance time difference is greater than a preset drift threshold, the global target judgment torque limit of the servo cover spindle is increased. When the channel friction resistance range is greater than or equal to the preset dispersion threshold, index positioning is performed based on the reference scale channel resistance vector to determine the target filling pump corresponding to the target index, and a back suction command is issued to the target filling pump. The system retrieves the en route packaging bottles being filled by the target filling pump from the transport trajectory station register, and when the en route packaging bottles arrive at the capping station, it issues a local torque reduction protection command for the en route packaging bottles to the servo capping spindle.

[0015] The present invention has the following beneficial effects: In this invention, the resistance value determination unit determines the equivalent capping resistance value of each bottle for each packaging bottle by the servo capping spindle, establishing the correlation between capping mechanical data and filling. The interference decoupling processing unit performs interference decoupling processing based on the equivalent capping resistance value of each bottle corresponding to each filling pump, separating the global time trend and local spatial differences from the coupled resistance data, eliminating the masking effect of product interference on anomaly judgment. Finally, the adaptive compensation control unit performs adaptive compensation control based on the global resistance vector of consumables and the single-channel friction resistance vector, adopting differentiated defense strategies for two different anomalies: global hardening and local liquid contamination. This effectively solves the problem of anomaly identification and defense in multi-head filling and capping production lines under the condition of data misalignment and product interference, reducing the probability of misjudgment shutdown and continuous bottle breakage. Attached Figure Description

[0016] Figure 1 This is a system architecture diagram of an intelligent control system for automatic filling and capping provided in one embodiment of the present invention; Figure 2 This is a flowchart illustrating an intelligent control method for automatic filling and capping, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent control system for automatic filling and capping based on the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific solution for an intelligent control system for automatic filling and capping provided by the present invention.

[0020] Please see Figure 1 The diagram illustrates a system architecture of an intelligent control system for automatic filling and capping according to an embodiment of the present invention. The intelligent control system 10 for automatic filling and capping includes: The resistance value determination unit 11 is used to determine the equivalent capping resistance value of each bottle for each packaged bottle by the servo capping spindle.

[0021] Each bottle is filled by its corresponding filling pump before being capped. The equivalent capping resistance value of a single bottle is used to characterize the average resistance of the servo capping spindle during a single capping process.

[0022] In production lines with multiple filling pumps operating in parallel, each bottle is filled by a specific filling pump and then transferred to the capping station via a star wheel conveyor. Due to differences in the sealing performance degradation of different filling pumps, leakage from a particular pump can cause abnormal frictional characteristics in the bottles along its corresponding channel during capping. Therefore, it is necessary to spatially link the mechanical data measured at the capping station with the upstream filling pumps to trace the source of the anomaly.

[0023] Optionally, the resistance value determination unit 11 includes: The entity status determination module is used to obtain the peak value of the downward current of the servo capping spindle in each single capping process, compare the peak value of the downward current with the preset idling current threshold, and determine the entity status of the capping station in each single capping process.

[0024] Among them, the entity state is used to characterize whether the packaging bottle exists at the capping station in the corresponding single capping process.

[0025] At the capping station, the first step is to confirm the presence of a real bottle. Because bottle height tolerances can lead to misjudgments, relying solely on fixed position sensors is insufficient to reliably identify the bottle's position. The system collects the current characteristics of the servo capping spindle during a single downward stroke and utilizes the positive correlation between the peak current and the mechanical load to determine whether the capping spindle is actually in contact with the bottle.

[0026] Specifically, the entity status determination module is used to: extract the maximum torque current value of multiple no-load pressing actions during the no-load trial operation of the equipment, and determine the preset no-load current threshold; if the peak pressing current is less than the preset no-load current threshold, determine that the entity status indicates the absence of packaging bottles, and generate a packaging bottle missing pulse signal, and perform a null value overwrite operation on the corresponding data bit in the conveyor trajectory station register in response to the packaging bottle missing pulse signal; if the peak pressing current is greater than or equal to the preset no-load current threshold, determine that the entity status indicates the presence of packaging bottles.

[0027] The system pre-calibrates a preset no-load current threshold through no-load operation. During the equipment trial run and ensuring that no bottles flow through the capping station, the system controls the servo capping spindle to continuously perform a preset number of no-load pressing and capping actions. For example, if the preset number is set to 100 times, the system extracts the maximum torque current value during these 100 no-load actions and adds a preset percentage (e.g., 5%) of tolerance margin to this maximum value. The sum of these values ​​is set as the preset no-load current threshold. This preset no-load current threshold represents the reference limit current required for the servo spindle to overcome its own mechanical transmission friction and gravity. Through the current threshold calibration during the no-load trial run, the system establishes a no-bottle no-load discrimination benchmark that matches the mechanical characteristics of the equipment, avoiding misjudgments caused by individual differences in the equipment.

[0028] Once continuous production begins, the system continuously collects the downward torque current along the Z-axis of the servo motor for each downward pressing stroke of the servo capping spindle. When the bottom of the pressing stroke is reached, the system extracts the data point with the highest value from the collected downward torque current timing data and records it as the peak value of the downward current. The system then compares the acquired peak value of the downward current with a pre-calibrated preset idling current threshold.

[0029] If the comparison result shows that the peak value of the down-pressure current is less than the preset idling current threshold, the system determines that the current capping spindle is not in contact with the packaging bottle. At this time, the system outputs a packaging bottle missing pulse signal and, in response to the packaging bottle missing pulse signal, checks the current record status of the corresponding data bit in the conveyor trajectory station register. If the status shows that the bottle is present, it is determined that the physical object is out of contact with the data queue. The system immediately performs a forced null value overwrite operation on the data bit, modifying its record to represent an invalid null position. The null position record is directly truncated and no longer transmitted to the downstream calculation stage, thereby completely avoiding the pollution of the subsequent matrix by the null data.

[0030] If the comparison result shows that the peak value of the down-voltage current is greater than or equal to the preset no-load current threshold, the system determines that there is a real packaging bottle at the current capping station and triggers the subsequent resistance calculation logic. Through dynamic comparison of the peak current and the no-load threshold, as well as the null value overwrite mechanism for missing pulse signals, the system corrects data misalignment caused by manual sampling or broken bottle rejection in real time, preventing no-load data from contaminating subsequent matrices.

[0031] The resistance value calculation module is used to obtain the rotational torque, rotational angle, and total downward displacement of the servo capping spindle during a single capping process when the physical state characterization involves a packaged bottle, and to determine the equivalent capping resistance value of a single bottle based on the rotational torque, rotational angle, and total downward displacement.

[0032] After confirming the presence of actual packaging bottles at the capping station, the system needs to further calculate the equivalent capping resistance value per bottle for this capping process.

[0033] Optionally, the resistance value calculation module is specifically used for: when the physical state characterizes the presence of a packaged bottle, obtaining the rotational torque, rotational angle, and total downward displacement of the servo capping spindle during a single capping process; when the total downward displacement is greater than a preset effective displacement threshold, calculating the equivalent capping resistance of a single bottle based on the rotational torque, rotational angle, and total downward displacement, and determining the equivalent capping resistance value of a single bottle; and generating a missing packaged bottle pulse signal when the total downward displacement is less than or equal to the preset effective displacement threshold.

[0034] When a bottle is detected, the system simultaneously extracts the real-time rotational torque and rotational angle output by the spindle during that single bottle capping cycle. To eliminate subsequent division failures caused by sensor zero values ​​due to mechanical jamming or unexpected termination of the spindle midway through the operation, the system first obtains the total downward displacement (i.e., Z-axis displacement) from the start point to the actual end point of the stroke, and determines whether this total downward displacement exceeds a preset effective displacement threshold. For example, this preset effective displacement threshold is set to 10% of the standard normal downward stroke length of the equipment.

[0035] If the total downward displacement is less than or equal to the preset effective displacement threshold, the system determines that the action is an abnormally short and invalid stroke, skips subsequent calculations, and generates a missing bottle pulse signal to discard the record. If the total downward displacement is greater than the preset effective displacement threshold, the system calculates the equivalent capping resistance value of a single bottle using real-time rotational torque and real-time rotational angle. in, This represents the equivalent capping resistance value for a single bottle. This refers to the rotational torque of the servo capping spindle during a single capping process. This refers to the rotation angle (in radians) of the servo capping spindle during a single capping process. This represents the total downward displacement. The definite integral of the real-time rotating torque with respect to the rotation angle is used to characterize the total energy consumed in this cap rotation action. The above calculation formula transforms the complex spatial helical torque into an equivalent resistance value reflecting the overall downward displacement by dividing the total energy consumed by the cap rotation by the total downward displacement. The calculation result... The larger the value, the higher the overall average resistance that the servo spindle overcomes during the process of tightening the plastic packaging cap into place.

[0036] The traceability writing module is used to trace the equivalent capping resistance value of a single bottle to the filling pump number based on the conveying trajectory station register, determine the corresponding filling pump number identifier, and write the equivalent capping resistance value of a single bottle to the channel buffer queue corresponding to the filling pump number identifier.

[0037] After obtaining the equivalent capping resistance value for a single bottle, the system needs to spatially align this value with the upstream filling pump. Since the production line uses a star-wheel conveyor to move the bottles between stations, each bottle undergoes multiple cycle steps from the filling station to the capping station. The system reads the station flow information recorded in the conveyor trajectory station register and traces the resistance value measured at the current capping station back to the source filling pump that performed the injection operation, thus establishing a correspondence between downstream mechanical data and the upstream filling pump number identifier. After tracing the source, the system writes the resistance value carrying the filling pump number identifier into the corresponding channel buffer queue, allowing the historical resistance data of the same filling pump channel to accumulate continuously in time.

[0038] Specifically, the traceability writing module is used to: extract the data index number of the packaging bottle corresponding to the single-bottle equivalent capping resistance value from the conveying trajectory station register; perform a reverse search in the historical data sequence of the conveying trajectory station register according to the data index number to locate the filling station record that performs the liquid injection process on the packaging bottle; extract the filling pump number identifier bound to the filling station record; and write the single-bottle equivalent capping resistance value of the packaging bottle into the channel buffer queue corresponding to the filling pump number identifier.

[0039] After successfully calculating the equivalent capping resistance value for a single bottle, the system extracts the unique data index number assigned to the current bottle from the conveyor trajectory station register. Using this data index number as a matching index, it performs a forward-backward search operation in the front-end historical work trajectory stored in the conveyor trajectory station register, searching back towards the source to the filling station record that triggered the filling process for this bottle. Through the reverse search of the data index number, the system locates the source filling station for each bottle, establishing a correspondence between downstream capping mechanical data and upstream filling equipment.

[0040] Next, the system extracts the filling pump number identifier bound to the filling station record, combines the equivalent capping resistance value of a single bottle with the filling pump number identifier, and generates a data object carrying the source filling pump number identifier. Based on the value in the source filling pump number identifier, the system writes the data object into a pre-divided independent channel buffer queue in the controller's memory, corresponding one-to-one with each filling pump station, for temporary storage and accumulation. After the data enters the buffer queue, the system assigns a time sequence number to the current independent sequence, thereby completing the spatial alignment task between the mechanical data of the downstream capping station and the specific upstream filling source.

[0041] The interference decoupling processing unit 12 is used to perform interference decoupling processing based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, and to generate a global resistance vector and a single-channel friction resistance vector for consumables.

[0042] Among them, the global resistance vector of consumables is used to characterize the overall resistance change trend of each filling pump under different time sliding batches, and the single-channel friction resistance vector is used to characterize the friction resistance difference of different filling pumps within the same time sliding batch.

[0043] During long-term continuous production, fluctuations in ambient temperature can cause a uniform change in the hardness of the entire batch of packaging materials. This global resistance drift can mask localized decreases in frictional resistance caused by dripping in individual channels. If a fixed threshold is directly set for the original resistance data, the rise in the global baseline will compress or even obscure the relative differences in local anomalies, causing the early warning mechanism to fail. Therefore, the system needs to perform interference decoupling processing on the accumulated resistance data of each channel, separating the coupled global temporal trend and local spatial differences into independent feature vectors, thereby eliminating the influence of product interference on anomaly detection.

[0044] Optionally, the interference decoupling processing unit 12 includes: The historical matrix construction module is used to update historical resistance data based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, and generate a historical capping resistance matrix.

[0045] The historical capping resistance matrix records the resistance characteristics of each filling pump channel within multiple consecutive time-sliding batches. The accumulated equivalent capping resistance values ​​per bottle in each channel's buffer queue are discrete one-dimensional time-series data. To analyze the resistance variation patterns of multiple filling pump channels within the same time dimension, it is necessary to align the discrete data of each channel to construct a unified two-dimensional data structure. In this invention, the historical capping resistance matrix uses time-sliding batches as rows and filling pump channels as columns, recording the representative resistance values ​​of each filling pump channel within multiple consecutive observation periods, thus achieving a unified representation of multi-channel data in the time dimension.

[0046] Specifically, the historical matrix construction module is used for: Whenever the cumulative number of cap-screwing actions reaches the preset sliding window size, the historical resistance data for the corresponding time-based sliding batch is updated. Each preset sliding window corresponds to one time-based sliding batch.

[0047] Before commencing formal production, the system acquires initial baseline data during the trial run phase using the first batch of qualified bottles entering the production line. Specifically, the system controls the filling pump, conveyor turntable, and servo capping spindle to operate at a normal production pace, processing the first N bottles (N is a preset value, for example, an integer multiple of the total number of elements in the historical capping resistance matrix, such as...). The complete filling and capping process is performed, and N resistance values ​​are obtained according to the aforementioned method for calculating the equivalent capping resistance of a single bottle. The arithmetic mean of the N resistance values ​​is taken, and this average value is set as the baseline work constant. This baseline work constant represents the normal resistance level of the current batch of consumables and under filling conditions, effectively avoiding the initial order-of-magnitude deviation introduced by the no-load value.

[0048] After obtaining the baseline work constant, the system constructs a two-dimensional data structure (i.e., the historical capping resistance matrix) in the controller's memory. The system sets the number of columns in this two-dimensional data structure to be... That is, the total number of filling pump channels, set the number of rows to... The number of time-sliding batches (e.g., set to 20 rows) yields the historical cap resistance matrix. Subsequently, the system will acquire the baseline work constant. Fill in the historical screw cap resistance matrix All In the initial cells, placeholders are used to initialize the matrix.

[0049] Once in production mode, the system activates a global sliding observation window within the controller, containing a preset total number of actions. This window is used to forcibly truncate the time period to align the data across each filling pump channel. The preset total number of actions is set to the total number of capping operations corresponding to three rotations of the star wheel turntable, for example, 300 times. The system continuously accumulates the effective downward pressing actions of the servo capping spindle. Whenever the accumulated number reaches the total number of capping operations, the system determines that the time-sliding batch has completed one forward step operation and proceeds to the next time-sliding batch.

[0050] Within the time-sliding batch, the channel buffer queue corresponding to each filling pump is traversed to obtain the equivalent capping resistance value of each bottle in the time-sliding batch for each filling pump channel. Based on the equivalent capping resistance value of each bottle in the time-sliding batch for each filling pump channel, the representative value of the single-channel resistance of each filling pump channel in the time-sliding batch is determined. Based on the representative value of the single-channel resistance of each filling pump channel in the time-sliding batch, the historical resistance data is updated to generate a historical capping resistance matrix.

[0051] As the time-sliding batch completes its stepping, the system sequentially traverses the search numbers from 1 to... The channel buffer queues corresponding to each filling pump. For the currently iterating... The system retrieves the first channel cache queue within the current time sliding batch. The number of valid samples accumulated in the current time-sliding batch is recorded as follows: All newly accumulated single-bottle equivalent capping resistance values ​​in the channel buffer queue. .

[0052] To prevent a division-to-zero error or computational crash from occurring because a large number of bottles corresponding to a specific channel at the front end are continuously removed, resulting in no samples available for calculation within the time-sliding batch, the system executes a zero-sample protection logic before calculation.

[0053] First, the system determines the number of... Number of valid samples in each channel buffer queue Is it equal to 0? When the result is... Furthermore, when the current time-sliding batch is the first time-sliding batch after the system enters the production state, the system extracts the pre-calibrated baseline work constant and assigns it to the first time-sliding batch. The single-channel resistance representative value of each channel buffer queue; when the judgment result is... Furthermore, if the current time is not the first time-sliding batch, the system reads the previous time-sliding batch corresponding to the first time-sliding batch from the historical cap resistance matrix. The historical single-channel resistance value is used as the reference value, and this historical single-channel resistance value is inherited and assigned to the current column. The single-channel resistance representative value of each channel buffer queue is used to ensure the numerical continuity of the historical capping resistance matrix and prevent subsequent iterations from failing to converge due to missing samples; when the judgment result is... At that time, the system extracts The sum of the equivalent capping resistance values ​​of each single bottle, divided by the sample size, is used as the representative value of the resistance of that single channel.

[0054] The system has completed all tasks. After processing the single-channel resistance representative value of each channel buffer queue, the obtained value will be... The single-channel resistance representative value is arranged according to the number. from to The order of these elements is used to horizontally combine and concatenate them into a complete row vector. The system then inserts this newly generated row vector as the latest row into the historical cap resistance matrix. At the very bottom. Simultaneously, the system removes this historical screw cap resistance matrix. The row vector data at the very top (i.e., the oldest row) is used to maintain the historical screw cap resistance matrix. The dimensions of two-dimensional rows and columns are absolutely constant.

[0055] The rank reduction feature decoupling module is used to perform rank reduction feature decoupling on the historical screw cap resistance matrix to generate a global resistance vector and a single-channel friction resistance vector for consumables.

[0056] The historical capping resistance matrix contains both global resistance drift caused by changes in ambient temperature over time and local friction differences between different filling pump channels due to individual equipment variations or drip anomalies. To separate these two types of features from the coupled data, the system performs rank reduction feature decoupling processing on the historical capping resistance matrix. By forcibly limiting the rank of the matrix to a lower dimension, the original matrix is ​​approximately reconstructed as a product of global temporal features and local spatial features, thus achieving data decoupling under product interference conditions.

[0057] Specifically, the rank reduction feature decoupling module is used to: perform non-negativity pre-constraint processing on the historical capping resistance matrix, assign preset bias values ​​to elements in the historical capping resistance matrix that do not meet the non-negativity condition constraints, and generate a non-negativity constraint matrix; and perform asynchronous factorization on the non-negativity constraint matrix through the rank reduction feature decoupling algorithm to generate a global resistance vector and a single-channel friction resistance vector for consumables.

[0058] After the historical cap resistance matrix is ​​updated, the system calls a pre-defined eigenvalue decomposition algorithm to decouple the product interference data. To avoid complex iterative algorithms blocking the millisecond-level response cycle of the underlying servo motor, the system controller deploys this eigenvalue decomposition process asynchronously in a separate background thread in the non-real-time application layer.

[0059] Because the rank-reduction feature decoupling algorithm requires that the input matrix strictly not contain negative numbers, negative equivalent drag values ​​may be recorded under the influence of zero-point drift of physical sensors or rare reverse interference. Therefore, the system performs a non-negativity pre-constraint check before inputting data. The system iterates through all data elements in the historical cap drag matrix. If an element with a value less than or equal to 0 is detected, the system forcibly assigns that non-positive element a preset minimum positive bias (e.g., set to 0.001). Through the non-negativity pre-constraint check, the system eliminates negative anomalies introduced by sensor zero-point drift or reverse interference, satisfying the strict non-negativity requirement of the rank-reduction feature decoupling algorithm for the input matrix.

[0060] After completing the nonnegativity constraint check, the system invokes the rank-reduced feature decoupling algorithm. The system sets the iteration stopping condition to be that the rate of change of the reconstructed residuals in two consecutive iterations is less than a preset rate of change (e.g., ...). This may occur when the maximum number of iterations is reached (e.g., set to 2000). To ignore random noise in the data and extract the dominant physical trend, the system limits the rank reduction parameter of this rank reduction feature decoupling algorithm to 1. The reconstructed residual can be defined as a nonnegative constraint matrix. With reconstruction matrix The sum of squares of the Euclidean distances between the two iterations, i.e. the square of the Frobenius norm, is used to calculate the rate of change of the reconstruction residuals in the current iteration relative to the reconstruction residuals in the previous iteration, which is the rate of change of the reconstruction residuals.

[0061] For example, the asynchronous factorization process includes: constructing a loss function based on Euclidean distance. ,in Represents a nonnegative constraint matrix. Represents the column vector generated iteratively. This represents the row vector generated iteratively; iterative convergence is achieved through a multiplicative update rule. This multiplicative update rule runs in a background thread of the controller, specifically by updating the vector vector sequentially along the time batch direction. Each element, and updated sequentially in vector form along the filling pump channel direction. Repeat each element until the preset convergence condition is met, and output the final value. As the global resistance vector of consumables, the final output is... As a single-channel frictional resistance vector. Furthermore, before entering the multiplicative iterative update, it is necessary to... and The elements are initialized randomly and non-negatively. Specifically, for Each element and Each element , and assign it a value as a random number that is uniformly distributed within the interval (0,1].

[0062] Furthermore, the multiplicative update rule is configured as follows: in, It is a dimension of The nonnegative constraint matrix, It is a dimension of The global drag vector of consumables, The first in the global drag vector of consumables Elements in a time-sliding batch. It is a dimension of The single-channel frictional resistance vector, The first in the single-channel frictional resistance vector The elements of each filling pump channel. When the rate of change of the reconstruction residual between two consecutive iterations is less than If the maximum number of iterations is reached, the preset convergence condition is determined to be met.

[0063] Based on the above configuration, the system performs factorization on the non-negative constraint matrix using a rank-reduced feature decoupling algorithm, approximately reconstructing the non-negative constraint matrix as a product of a single column vector and a single row vector. The system extracts the single column vector as the global resistance vector for the consumables. This global resistance vector for consumables includes... For each row element corresponding to a time-sliding batch, since the rank of the reduced dimension is limited to 1, the numerical trend of the elements of the global resistance vector of this consumable excludes the spatial differences between each filling pump channel, representing the continuous... The single-time evolution trend of the plastic packaging cap consumables hardening due to the decrease in ambient temperature faced by the whole machine under the sliding batch of time.

[0064] The system synchronously extracts a single-row vector as a single-channel friction resistance vector. This single-channel friction resistance vector includes... The column elements corresponding to the filling pump, the element values ​​of this single-channel friction resistance vector exclude the influence of overall time drift. If the specific local element values ​​are extremely small, it indicates that the packaging bottle filled by the corresponding filling pump has liquid stringing and dripping, resulting in excessive lubrication effect on its external threads at the capping station.

[0065] The adaptive compensation control unit 13 is used to perform adaptive compensation control based on the global resistance vector of the consumable and the single-channel frictional resistance vector.

[0066] After obtaining the global resistance trend and local friction differences after decoupling, the system adopts differentiated compensation strategies for different types of abnormal operating conditions. When the global resistance shows an overall upward trend, it indicates that the ambient temperature has decreased, causing the consumables to harden. In this case, the judgment criterion for capping torque needs to be adjusted accordingly to avoid misjudging normal capping resistance as overload. When the frictional resistance of a specific channel is significantly lower than that of other channels, it indicates that there is liquid lubrication due to dripping in that channel. In this case, targeted intervention actions need to be taken on the source filling pump of that channel and the affected packaging bottles in transit to prevent bottle breakage due to over-twisting.

[0067] Optionally, the adaptive compensation control unit 13 includes: The scaling and index generation module is used to perform benchmark scaling on the global resistance vector and single-channel friction resistance vector of consumables, respectively, to generate a normalized consumable resistance vector and a benchmark-scaled channel resistance vector. Based on the normalized consumable resistance vector, the module determines the consumable resistance time difference, and based on the benchmark-scaled channel resistance vector, the module determines the channel friction resistance range.

[0068] Among them, the consumable resistance time difference is used to characterize the overall resistance deviation of the whole machine consumables under continuous time sliding batch, and the channel friction resistance range is used to characterize the degree of dispersion of friction resistance corresponding to different filling pumps within the same time sliding batch.

[0069] Because the rank reduction feature decoupling algorithm suffers from scale indeterminacy in its mathematical iteration principle—meaning that the iteration results may produce arbitrary relative scaling factors, causing the output vector to lose its true physical drag magnitude—this random scaling is eliminated by extracting the mean and performing a benchmark scaling mapping with inverse multiplication and division. This restores the decomposed feature vector to the same physical order of magnitude space as the initial measured data.

[0070] The arithmetic mean of all elements in the global resistance vector of consumables is extracted as a scaling factor. A division operation is performed on the global resistance vector to complete the scaling mapping; for example, each element in the global resistance vector is divided by the scaling factor confirmed above. The resulting column vector is used as the normalized consumable resistance vector. The generated normalized consumable resistance vector removes the interference of the multiplication coefficients, and the fluctuations of its internal elements represent the relative proportional evolution of the baseline resistance over time.

[0071] For a single-channel friction resistance vector, a reverse multiplication compensation operation is performed to complete the scaling mapping. For example, each element in the single-channel friction resistance vector is multiplied by a scaling factor to obtain a baseline scaling channel resistance vector. After this scaling mapping process, the numerical magnitude of the elements inside the baseline scaling channel resistance vector is remapped back to the same physical order of magnitude as the initial underlying measured mechanical data.

[0072] After scaling, the time difference representing environmental drift is extracted from the normalized consumable resistance vector. To enhance the system's robustness against single-sensor noise, a block smoothing extraction logic is used. The first three consecutive elements (the oldest elements in the time sequence) of the normalized consumable resistance vector are extracted, and their arithmetic mean is calculated and recorded as the initial smoothing mean. Simultaneously, the last three consecutive elements (the newest elements in the time sequence) are extracted, and their arithmetic mean is calculated and recorded as the final smoothing mean. Subtracting the initial smoothing mean from the final smoothing mean yields the consumable resistance time difference. If the consumable resistance time difference is greater than zero and continues to rise, it indicates that the material of the entire batch of plastic packaging caps is hardening as the ambient temperature decreases. The consumable resistance time difference is a dimensionless proportional difference, representing the relative change in normalized resistance at the beginning and end of the current monitoring window.

[0073] Subsequently, the channel friction resistance range, representing local anomalies, is extracted from the baseline-scaled channel resistance vector. The system iterates through all channels in the baseline-scaled channel resistance vector. For each column of elements, the element with the smallest value is extracted and recorded as the minimum single-channel resistance value, and the element with the largest value is extracted and recorded as the maximum single-channel resistance value. Then, the minimum single-channel resistance value is subtracted from the maximum single-channel resistance value to obtain the channel friction resistance range. The larger the value of this channel friction resistance range, the greater the dispersion of frictional forces between multiple filling pump channels, and the higher the probability that a specific filling pump will leak fluid, causing abnormal lubrication of the external threads of the packaging bottles passing through that station.

[0074] The compensation control execution module is used to perform adaptive compensation control based on the time difference of consumable resistance and the range of channel friction resistance.

[0075] Specifically, the compensation control execution module is used for: When the time difference of consumable resistance exceeds the preset drift threshold, the upper limit of the global target judgment torque of the servo cover spindle is increased.

[0076] When the channel friction resistance range is greater than or equal to the preset dispersion threshold, index positioning is performed based on the benchmark scale channel resistance vector to determine the target filling pump corresponding to the target index, and a back suction command is sent to the target filling pump. The system retrieves the en route packaging bottles being filled by the target filling pump from the transport trajectory station register, and sends a local torque reduction protection command to the servo capping spindle when the en route packaging bottles arrive at the capping station.

[0077] The global compensation and local protection mentioned above are independent and parallel, and can be triggered and executed simultaneously within the same sliding batch. When the system starts up and is put into continuous production, in order to prevent false large differences between the reference work constant and the actual production data, a matrix preheating flag is set in the controller to accumulate the total number of batches that have completed the stepping in real time and determine whether the total number of batches is greater than or equal to the number of rows in the historical capping resistance matrix.

[0078] When the total number of batches is less than the number of rows, it is determined that the current historical capping resistance matrix has not yet been completely covered by the actual production data. At this time, the matrix preheating flag remains locked. In the locked state, regardless of the received consumable resistance time difference or channel friction resistance range value, no torque compensation or pull-back commands are output, and the servo capping spindle executes the action according to the default parameters. When the total number of batches is greater than or equal to the number of rows, it is determined that the actual single-channel resistance representative value has completely rolled over the initial constant. At this time, the system switches the matrix preheating flag to the unlocked state, allowing the system to perform subsequent difference comparison and closed-loop compensation operations.

[0079] In the unlocked state, the system first retrieves the preset dispersion threshold and preset drift threshold pre-set in the controller. For example, the preset dispersion threshold is obtained as follows: during the drying trial operation phase when the equipment has no liquid added, the system records the maximum historical value of the range of the representative values ​​of the resistance of each channel within multiple consecutive sliding observation windows, and multiplies the maximum historical value by a preset safety margin constant (for example, a value of 1.3) to obtain the preset dispersion threshold.

[0080] The preset drift threshold is based on the thermoplasticity parameters of the packaging material in the current batch. The preset drift threshold is a dimensionless proportional threshold with dimensions consistent with the time difference of the material's resistance. Specific preset steps include: during the offline equipment debugging phase, using a constant temperature chamber to simulate multiple temperature points covering the expected production environment temperature range, performing capping experiments on dry empty bottles with the same batch of packaging caps, obtaining the tightening torque at different temperatures, and plotting a temperature-torque characteristic curve; extracting the temperature inflection point on this characteristic curve where the torque-temperature change rate first exceeds a preset sensitivity threshold (e.g., -1.5% / ℃), determining the normalized resistance change rate at this temperature inflection point, and first calculating the corresponding equivalent capping resistance value for each temperature point using the formula for calculating the equivalent capping resistance value of a single bottle. The resistance values ​​are collected and arranged in descending order of temperature to form a reference resistance vector. The arithmetic mean of all elements in the reference resistance vector is calculated. Each element in the reference resistance vector is divided by this arithmetic mean to generate a normalized reference resistance vector. The difference between the last and first normalized resistance values ​​in the normalized reference resistance vector is calculated and used as the reference drift sensitivity (if the difference is less than 0, it is set to 0). The reference drift sensitivity is multiplied by a preset safety factor (e.g., 1.2) to obtain a preset drift threshold. This threshold is used to determine whether overall hardening drift of the consumables has occurred due to a decrease in ambient temperature.

[0081] The system compares the currently acquired channel friction resistance range with a preset dispersion threshold. When the channel friction resistance range is less than the preset dispersion threshold, it is determined that the frictional differences between all filling pump channels are within the normal tolerance range, and no abnormal liquid stringing or dripping has occurred. At this point, it further checks whether the consumable resistance time difference is greater than a preset drift threshold. If the consumable resistance time difference is greater than this preset drift threshold, it is confirmed that the overall hardness of the consumables has increased due to environmental cooling. The acquired consumable resistance time difference is multiplied by a preset conversion constant to obtain a quantitative torque compensation increment. A torque compensation action command containing this torque compensation increment is sent to the underlying driver of the servo capping spindle. This torque compensation action command instructs the global target torque limit of the current servo capping spindle to be synchronously increased. Through this increase, the system prevents plastic packaging caps that are originally of acceptable quality but harden at low temperatures from being incorrectly judged as capping overloads, thereby eliminating false alarms and unplanned shutdowns.

[0082] The conversion constant is predetermined through calibration experiments. The determination steps include: operating the production line under different ambient temperature conditions with the equipment unloaded and using the same batch of packaging materials; continuously recording the normalized consumable resistance vector for each time-sliding batch and calculating the corresponding consumable resistance time difference; simultaneously, for each temperature condition, determining the torque compensation setpoint required to ensure the capping qualification rate through process experiments, and establishing a data set of consumable resistance time difference versus torque compensation; using the least squares method to perform linear regression fitting on the data set, and using the slope of the resulting regression line as the conversion constant. The dimension of this conversion constant is torque units, and its physical meaning is the torque compensation increment corresponding to each unit change in the consumable resistance time difference.

[0083] When the range of frictional resistance between channels is greater than or equal to a preset dispersion threshold, it is determined that liquid-lubricated differentiation has occurred between multiple filling pump channels. The specific column index that generates the minimum single-channel resistance value is retrieved from the baseline scaling channel resistance vector, and the filling pump number identifier bound to that column index is extracted as the target filling pump.

[0084] After identifying the target filling pump that caused the anomaly, a source back-suction command is sent to the target filling pump via the production line communication bus. At the moment of liquid cut-off after the piston completes its stroke in a subsequent single injection cycle, a set number of pulses (e.g., 50 pulses) are sent to the stepper motor controlling the piston to force the piston back, causing the residual liquid level at the nozzle to quickly shrink, thereby directly cutting off the source of contamination from the continuous stringing and dripping.

[0085] Secondly, for bottles that have already been contaminated with medication and are moving towards the capping station on the conveyor turntable, the location markers of all en route bottles bearing the same target filling pump number are read and retrieved from the conveyor trajectory station register. When the sensor detects an en route bottle with this marker stepping directly below the servo capping spindle, a local torque reduction protection command is issued to the servo capping spindle for this single action, instructing the target torque limit for that en route bottle to be lowered to a preset low percentage of the conventional value benchmark (e.g., reduced to 35% of the conventional value benchmark). Because the frictional resistance of the bottle's external threads drops sharply when contaminated with medication, the servo capping spindle does not encounter the expected mechanical resistance during the downward and rotational processes at the conventionally set higher torque limit, resulting in uncontrolled over-rotation and abnormal continuous downward pressure, ultimately crushing the bottle's neck. By lowering the upper limit of the target torque, when the servo capping spindle encounters an abnormally smooth bottleneck, it can immediately determine that it has been tightened when it reaches the lower torque limit, and stop applying force and pressing in advance, thereby preventing the accident of over-tightening and bottle breakage.

[0086] After the servo capping spindle completes a single torque reduction operation, a safety recovery mechanism for the target torque is executed for the next bottle that is about to flow into the capping station, in order to prevent instruction overwrite conflicts caused by multiple consecutive liquid-soaked bottles.

[0087] Before the servo capping spindle performs a pressing action on the next in-transit bottle, it re-detects the data identifier corresponding to the next bottle in the conveyor trajectory station register and determines whether the filling pump number recorded in the identifier is the same as the target filling pump that caused the over-limit.

[0088] If the data identifier of the next bottle still contains the target filling pump number, indicating a continuous occurrence of contaminated bottles from the same source, no recovery action is performed, and the local torque reduction protection command continues. Only when it is verified that the data identifier of the next bottle does not contain the target filling pump number will the system issue a recovery command to the servo capping spindle. This recovery command restores the target judgment torque upper limit from the reduced state to the global normal reference value before adjustment, ensuring normal capping judgment for subsequent normal bottles.

[0089] Meanwhile, after issuing a back suction command to the target filling pump, the data record of the corresponding data bit of the current capping station in the conveying trajectory station register is continuously monitored; when the null value of no packaging bottle is detected for the data bit for a preset number of consecutive times (for example, twice in a row), it indicates that all packaging bottles in transit contaminated by the target filling pump have passed the capping station. At this time, the local torque reduction protection command is forcibly canceled and the global normal target judgment torque limit is restored.

[0090] Please see Figure 2 The diagram illustrates a method flowchart of an intelligent control method for automatic filling and capping according to an embodiment of the present invention. The method includes the following steps: S100. Determine the equivalent capping resistance value of each bottle for each packaged bottle by the servo capping spindle. Each packaged bottle is filled by its corresponding filling pump before capping. The equivalent capping resistance value of each bottle is used to characterize the average resistance of the servo capping spindle in a single capping process. S200. Based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, interference decoupling processing is performed to generate a global resistance vector and a single-channel friction resistance vector for consumables. The global resistance vector of consumables is used to characterize the overall resistance change trend of each filling pump under different time sliding batches, and the single-channel friction resistance vector is used to characterize the friction resistance difference of different filling pumps within the same time sliding batch. S300 performs adaptive compensation control based on the global resistance vector of the consumables and the single-channel frictional resistance vector.

[0091] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. An intelligent control system for automatic filling and capping, characterized in that, include: The resistance value determination unit is used to determine the equivalent capping resistance value of a single bottle for each packaging bottle by the servo capping spindle; each packaging bottle is filled by its corresponding filling pump before capping, and the equivalent capping resistance value of a single bottle is used to characterize the average resistance of the servo capping spindle in a single capping process. The interference decoupling processing unit is used to perform interference decoupling processing based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, and generate a global resistance vector and a single-channel friction resistance vector for consumables. The global resistance vector of consumables is used to characterize the overall resistance change trend of each filling pump under different time sliding batches, and the single-channel friction resistance vector is used to characterize the friction resistance difference of different filling pumps within the same time sliding batch. An adaptive compensation control unit is used to perform adaptive compensation control based on the global resistance vector of the consumable and the single-channel frictional resistance vector.

2. The intelligent control system for automatic filling and capping according to claim 1, characterized in that, The resistance value determination unit includes: The entity state determination module is used to obtain the peak value of the downward current of the servo capping spindle in each single capping process, compare the peak value of the downward current with the preset idling current threshold, and determine the entity state of the capping station in each single capping process; the entity state is used to characterize whether there is a packaging bottle at the capping station in the corresponding single capping process. The resistance value calculation module is used to obtain the rotational torque, rotational angle, and total downward displacement of the servo capping spindle in the single capping process when the physical state characterizes the presence of a packaged bottle, and to determine the equivalent capping resistance value of a single bottle based on the rotational torque, the rotational angle, and the total downward displacement. The traceability writing module is used to trace the equivalent capping resistance value of a single bottle to the filling pump number based on the conveying trajectory station register, determine the corresponding filling pump number identifier, and write the equivalent capping resistance value of a single bottle to the channel buffer queue corresponding to the filling pump number identifier.

3. The intelligent control system for automatic filling and capping according to claim 2, characterized in that, The entity state determination module is specifically used for: During the no-load trial operation of the equipment, the maximum torque current value of multiple no-load pressure actions is extracted to determine the preset no-load current threshold. If the peak value of the down-pressure current is less than the preset idling current threshold, it is determined that the physical state characterization does not contain a packaging bottle, and a packaging bottle missing pulse signal is generated. In response to the packaging bottle missing pulse signal, a null value overwrite operation is performed on the corresponding data bit in the conveying trajectory station register. If the peak value of the down-current is greater than or equal to the preset idling current threshold, it is determined that the physical state characterizes the presence of a packaging bottle.

4. The intelligent control system for automatic filling and capping according to claim 2, characterized in that, The resistance value calculation module is specifically used for: When the physical state characterizes the presence of a packaged bottle, the rotational torque, rotational angle, and total downward displacement of the servo capping spindle are obtained during the single capping process. When the total downward displacement is greater than the preset effective displacement threshold, the equivalent capping resistance of a single bottle is calculated based on the rotational torque, the rotational angle, and the total downward displacement to determine the equivalent capping resistance value of a single bottle. When the total downward displacement is less than or equal to the preset effective displacement threshold, a packaging bottle missing pulse signal is generated.

5. The intelligent control system for automatic filling and capping according to claim 2, characterized in that, The source tracing writing module is specifically used for: Extract the data index number of the packaging bottle corresponding to the single bottle equivalent capping resistance value from the conveying trajectory station register, and perform a reverse search in the historical data sequence of the conveying trajectory station register according to the data index number to locate the filling station record that performs the liquid injection process on the packaging bottle; Extract the filling pump number identifier bound to the filling station record, and write the single-bottle equivalent capping resistance value of the packaging bottle into the channel buffer queue corresponding to the filling pump number identifier.

6. The intelligent control system for automatic filling and capping according to claim 1, characterized in that, The interference decoupling processing unit includes: The historical matrix construction module is used to update historical resistance data based on the equivalent capping resistance value of a single bottle corresponding to each filling pump, and generate a historical capping resistance matrix; the historical capping resistance matrix is ​​used to record the resistance characteristic data of each filling pump channel within multiple consecutive time sliding batches. The rank reduction feature decoupling module is used to perform rank reduction feature decoupling on the historical capping resistance matrix to generate the global resistance vector of the consumable and the single-channel friction resistance vector.

7. The intelligent control system for automatic filling and capping according to claim 6, characterized in that, The historical matrix construction module is specifically used for: Whenever the cumulative number of capping actions reaches the preset sliding window size, the historical resistance data of the corresponding time sliding batch is updated. Within the time-sliding batch, the channel buffer queue corresponding to each filling pump is traversed to obtain the equivalent capping resistance value of each bottle in the time-sliding batch for each filling pump channel; The representative value of the single-channel resistance of each filling pump channel in the time-sliding batch is determined based on the equivalent capping resistance value of all single bottles in each filling pump channel in the time-sliding batch. Historical resistance data is updated based on the single-channel resistance representative value of each filling pump channel in the time-sliding batch, generating a historical capping resistance matrix.

8. The intelligent control system for automatic filling and capping according to claim 6, characterized in that, The rank reduction feature decoupling module is specifically used for: Perform non-negativity pre-constraint processing on the historical screw cap resistance matrix, assign a preset bias value to the elements in the historical screw cap resistance matrix that do not meet the non-negativity condition constraint, and generate a non-negativity constraint matrix. The non-negative constraint matrix is ​​asynchronously factored using a rank-reducing feature decoupling algorithm to generate the global resistance vector of the consumables and the single-channel friction resistance vector.

9. The intelligent control system for automatic filling and capping according to claim 1, characterized in that, The adaptive compensation control unit includes: The scaling and index generation module is used to perform benchmark scaling on the global resistance vector of the consumables and the single-channel frictional resistance vector respectively, to generate a normalized consumables resistance vector and a benchmark-scaled channel resistance vector, and to determine the consumables resistance time difference based on the normalized consumables resistance vector, and to determine the channel frictional resistance range based on the benchmark-scaled channel resistance vector; the consumables resistance time difference is used to characterize the overall resistance offset of the whole machine consumables under continuous time sliding batch; the channel frictional resistance range is used to characterize the degree of dispersion of frictional resistance corresponding to different filling pumps within the same time sliding batch. The compensation control execution module is used to perform adaptive compensation control based on the time difference of the consumable resistance and the range of the channel friction resistance.

10. The intelligent control system for automatic filling and capping according to claim 9, characterized in that, The compensation control execution module is specifically used for: When the time difference of the consumable resistance is greater than the preset drift threshold, the upper limit of the global target judgment torque of the servo cover spindle is increased. When the channel friction resistance range is greater than or equal to the preset dispersion threshold, index positioning is performed based on the reference scale channel resistance vector to determine the target filling pump corresponding to the target index, and a back suction command is issued to the target filling pump. The system retrieves the en route packaging bottles being filled by the target filling pump from the transport trajectory station register, and when the en route packaging bottles arrive at the capping station, it issues a local torque reduction protection command for the en route packaging bottles to the servo capping spindle.

Images