A kind of anti-dripping adaptive fluid dispensing control method and system

Abstract

The present application relates to the technical field of fluid control, and discloses a kind of drip-proof adaptive fluid dispensing control method and system.The method includes obtaining image sequence, pressure data and viscosity variation data when fluid is disconnected;Positioning abnormal disconnection time, extracting the corresponding picture and extracting necking diameter and necking length therefrom;Based on necking parameter, calculate the real-time change rate of necking diameter, obtain the state identifier that fluid is about to be disconnected and predict the disconnection time;Integrate the comprehensive data set obtained by disconnection time prediction value and viscosity data, calibrate core parameters to generate adaptive signal waveform;Real-time flow rate data is obtained to correct waveform, and an optimized control signal is generated;Extract key intervention point to apply pulse intervention, calculate colloid residue, and if it exceeds the preset threshold, iterate optimization, otherwise, no intervention is needed, and finally a drip-free control sequence is obtained.This method can accurately capture the necking characteristics of the fluid, dynamically adapt to the working conditions, achieve drip-free dispensing, and improve control accuracy and stability.

Description

Technical Field

[0001] This invention relates to the field of fluid control technology, and in particular to an anti-drip adaptive fluid dispensing control method and system. Background Technology

[0002] With the rapid development of industrial automation and intelligence, fluid control technology, as a core component of industrial control systems for achieving precise fluid transport and regulation, has become a core supporting field in modern industrial manufacturing, widely used in key industries such as electronic packaging, medical device manufacturing, and precision assembly. Its control precision and stability directly determine the consistency of product quality and the potential for improving production efficiency. Furthermore, it is a crucial manifestation of the refined and intelligent regulation of industrial control systems, playing a key role in promoting the high-quality development of related industries.

[0003] Currently, most fluid dispensing control technologies in industrial control systems rely on preset fixed control parameters or simple feedback adjustment mechanisms. The core idea is to achieve fluid delivery and disconnection control by pre-setting static parameters such as flow rate and pressure, or by making coarse adjustments based on data from a single sensor. For example, some dispensing modules in industrial control systems only set fixed control signal waveforms based on the initial viscosity of the fluid, lacking the ability to adapt to dynamic changes in the fluid's state. Other systems, while introducing feedback mechanisms, only focus on single indicators such as pressure or flow rate, failing to conduct targeted analysis of key morphological characteristics during the fluid disconnection process. This results in the control precision of industrial control systems failing to meet the demands of high-end manufacturing.

[0004] Existing technologies cannot accurately capture the necking characteristics during fluid disconnection. The necking characteristic is the morphological change of the elongated connecting portion formed during fluid separation, involving key parameters such as necking diameter and length, which directly determine whether the fluid can be cleanly and efficiently disconnected. Due to the lack of real-time perception and analysis of this characteristic, existing technologies result in deviations between the control commands output by the industrial control system and the actual fluid state. At critical points where fluid disconnection is necessary, appropriate control actions cannot be applied, inevitably leading to fluid leakage or residue problems. This deficiency not only makes it difficult to improve control accuracy but also directly affects the reliability of industrial control systems in precision manufacturing scenarios. In other words, the core problem of existing technologies lies in the inability to accurately perceive the necking characteristics during fluid disconnection, making it difficult to accurately determine the timing of fluid disconnection and apply appropriate control actions, ultimately leading to unavoidable leakage or residue problems in the fluid dispensing stage of industrial control systems. Summary of the Invention

[0005] This invention provides an anti-drip adaptive fluid dispensing control method and system to solve the problem of dripping or residue caused by the inability to accurately capture the necking characteristics of fluid disconnection in industrial control systems.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides an anti-drip adaptive fluid dispensing control method, comprising: Acquire image sequences, pressure data, and viscosity change data of the fluid during the disconnection process; Based on the image sequence and pressure data, locate the abnormal disconnection moment, extract the image corresponding to the abnormal disconnection moment, and extract the necking diameter and necking length from it; The real-time rate of change of the necking diameter is calculated based on the necking diameter and the necking length. The state indicator that the fluid is about to be disconnected is obtained according to the necking length and the real-time rate of change. The timing of disconnection is predicted to obtain the predicted value of disconnection timing. The predicted disconnection timing value and the viscosity change data are integrated to obtain a comprehensive dataset. Core parameters related to signal amplitude and timing are extracted from the comprehensive dataset and calibrated. Based on the calibrated parameters, a signal waveform adapted to the fluid disconnection characteristics is generated. Acquire real-time flow velocity data, perform targeted correction on the signal waveform based on the real-time flow velocity data, and generate an optimized control signal based on the corrected signal waveform; Extract the key intervention points in the optimized control signal, apply pulse intervention according to the triggering sequence corresponding to the key intervention points, and then calculate the estimated value of the colloidal residue. If the estimated value exceeds the preset residual threshold, an iterative optimization process is performed. If the estimated value does not exceed the preset residual threshold, no further intervention is required, and a leak-free control sequence is finally obtained.

[0007] In one optional implementation, acquiring the image sequence, pressure data, and viscosity change data of the fluid during the disconnection process includes: Real-time data acquisition of the fluid disconnection process is performed using a sensor array to obtain image sequences and pressure data. High-precision sensors are used to monitor minute fluctuations in fluid viscosity in real time and obtain viscosity change data.

[0008] In one optional implementation, the step of locating the abnormal disconnection moment based on the image sequence and pressure data, extracting the image corresponding to the abnormal disconnection moment, and extracting the necking diameter and necking length from it includes: The image sequence is denoised and contrast enhanced, and then a multi-frame boundary point set of the necking region is obtained through edge detection; The multi-frame boundary point set is subjected to temporal tracking and inter-frame matching to extract the parameters of the necking morphology changing over time and determine the dynamic characteristics of the necking morphology. The dynamic features and the pressure data are aligned and fused along the time axis to obtain the correlation features between the neck constriction morphology and pressure changes. If the fluctuation amplitude of the associated feature is higher than the preset amplitude threshold, it is determined to be an abnormal disconnection state and the abnormal disconnection time is located. The image corresponding to the abnormal disconnection time is extracted and enlarged. The necking diameter and necking length are extracted from the enlarged image.

[0009] In one optional implementation, the step of calculating the real-time rate of change of the necking diameter based on the necking diameter and the necking length, obtaining a state indicator indicating that the fluid is about to disconnect based on the necking length and the real-time rate of change, and predicting the disconnection timing to obtain a predicted disconnection timing value includes: The necking diameter and the necking length are standardized to obtain a set of standard parameters; Based on the set of standard parameters, the real-time change rate of the necking diameter is calculated. If the necking diameter is less than a preset necking diameter threshold and the real-time change rate shows an increasing trend, the fluid state is determined to be about to disconnect, and the corresponding state identifier is obtained. Based on the status identifier, a time series analysis is performed on the standard parameter set to obtain the predicted disconnection timing value.

[0010] In one optional implementation, the process of integrating the predicted disconnection timing value with the viscosity change data to obtain a comprehensive dataset, extracting and calibrating core parameters related to signal amplitude and timing from the comprehensive dataset, and generating a signal waveform adapted to the fluid disconnection characteristics based on the calibrated parameters includes: By integrating the predicted disconnection timing with the viscosity change data, a comprehensive dataset is obtained; Dynamic simulation analysis is performed on the comprehensive dataset to extract core parameters related to signal amplitude and timing, thereby obtaining dynamic characteristic values; wherein, the dynamic simulation analysis is performed using a pre-established fluid database matching method; If the dynamic characteristic value exceeds the preset dynamic characteristic value threshold, the signal amplitude and timing are calibrated to obtain a combination of calibration parameters; Based on the combination of calibration parameters, a signal waveform adapted to the fluid disconnection characteristics is generated.

[0011] In one optional implementation, the steps of acquiring real-time flow velocity data, specifically correcting the signal waveform based on the real-time flow velocity data, and generating an optimized control signal based on the corrected signal waveform include: Real-time flow velocity data is collected by a flow sensor and combined with a preset flow velocity threshold to determine whether the flow velocity exceeds the limit. If it does not exceed the limit, the signal waveform is used as the correction waveform; if it exceeds the limit, it is marked as the waveform to be corrected. Based on the degree of deviation of the real-time flow rate data from the preset flow rate threshold, the amplitude and timing parameters of the waveform to be corrected are specifically modified to obtain the corrected waveform. Based on the corrected waveform, an optimized control signal is generated.

[0012] In one optional implementation, the step of extracting key intervention points from the optimized control signal, inputting the key intervention points into the actuator system, applying pulse intervention when the necking characteristic reaches its peak according to the triggering sequence corresponding to the key intervention points, and calculating the estimated colloidal residue amount includes: Extract key intervention points that are time-series correlated with the peak value of the necking characteristic from the optimized control signal. If the amplitude of the key intervention point does not meet the preset amplitude threshold, then re-extract the key intervention point and verify it. Obtain the trigger timing corresponding to the key intervention point, apply pulse intervention according to the trigger timing, and then calculate the estimated value of the colloidal residue.

[0013] In one optional implementation, the iterative optimization process includes: The records and results of this intervention are obtained and input into a preset neural network model. A set of optimized intervention parameters are output and the intervention is performed again. The updated estimate of the amount of colloid residue after the intervention is calculated. The process of obtaining the records and results of this intervention and inputting them into a preset neural network model is repeated, outputting a set of optimized intervention parameter combinations and performing the intervention again, and calculating the updated estimate of the colloid residue after the intervention, continues until the updated estimate is lower than the preset residue threshold.

[0014] Secondly, the present invention provides an anti-drip adaptive fluid dispensing control system, comprising: The data acquisition module is used to acquire image sequences, pressure data, and viscosity change data of the fluid during the disconnection process; The feature extraction module is used to locate the abnormal disconnection time based on the image sequence and pressure data, extract the image corresponding to the abnormal disconnection time, and extract the necking diameter and necking length from it; The disconnection prediction module is used to calculate the real-time change rate of the necking diameter based on the necking diameter and the necking length, obtain the state indicator that the fluid is about to disconnect based on the necking length and the real-time change rate, and predict the disconnection timing to obtain the disconnection timing prediction value. The waveform generation module is used to integrate the predicted disconnection timing value and the viscosity change data to obtain a comprehensive dataset. It extracts and calibrates the core parameters related to signal amplitude and timing from the comprehensive dataset and generates a signal waveform adapted to the fluid disconnection characteristics based on the calibrated parameters. The feedback optimization module is used to acquire real-time flow velocity data, perform targeted correction on the signal waveform based on the real-time flow velocity data, and generate an optimized control signal based on the corrected signal waveform. The pulse intervention module is used to extract key intervention points in the optimized control signal, apply pulse intervention according to the triggering sequence corresponding to the key intervention points, and then calculate the estimated value of the colloidal residue. The iterative calibration module is used to perform an iterative optimization process if the estimated value exceeds the preset residual threshold, and if the estimated value does not exceed the preset residual threshold, no further intervention is required, ultimately obtaining a leak-free control sequence.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention acquires image sequences and pressure data of fluid disconnection through a sensor array, extracts necking diameter and length parameters, locates the abnormal disconnection time, and generates an adaptive signal waveform by combining viscosity data. This multi-dimensional data acquisition accurately analyzes necking characteristics and enables precise prediction of disconnection timing, which can reduce the risk of leakage and improve control accuracy.

[0016] (2) This invention constructs a feedback loop, corrects the signal waveform according to the flow rate deviation, applies pulse intervention at the necking peak, and iteratively optimizes the parameters if the residual amount exceeds the standard, so that it can adapt to the working conditions in real time, accurately suppress dripping, reduce the residual amount, and improve the dispensing stability and product consistency.

[0017] (3) By constructing a closed-loop system for the entire process, the present invention integrates multi-source data into the control process, iteratively optimizes and corrects parameters, breaks through data limitations, can adapt to different fluids and working conditions, solves the leakage problem, and meets the needs of precision scenarios such as electronic packaging and medical device manufacturing. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a drip-proof adaptive fluid dispensing control method provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of an anti-drip adaptive fluid dispensing control system provided in the second embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 The first embodiment of the present invention provides an anti-drip adaptive fluid dispensing control method, comprising the following steps: S11, acquire image sequences, pressure data, and viscosity change data of the fluid during the disconnection process; S12, based on the image sequence and pressure data, locate the abnormal disconnection time, extract the image corresponding to the abnormal disconnection time, and extract the necking diameter and necking length from it; S13, calculate the real-time change rate of the necking diameter based on the necking diameter and the necking length, obtain the state indicator that the fluid is about to be disconnected according to the necking length and the real-time change rate, and predict the disconnection timing to obtain the disconnection timing prediction value. S14, integrate the predicted disconnection timing value with the viscosity change data to obtain a comprehensive dataset, extract and calibrate the core parameters related to signal amplitude and timing from the comprehensive dataset, and generate a signal waveform adapted to the fluid disconnection characteristics based on the calibrated parameters; S15, acquire real-time flow velocity data, perform targeted correction on the signal waveform based on the real-time flow velocity data, and generate an optimized control signal based on the corrected signal waveform; S16, extract the key intervention points in the optimized control signal, apply pulse intervention according to the triggering timing corresponding to the key intervention points, and then calculate the estimated value of the colloid residue; S17, if the estimated value exceeds the preset residual threshold, an iterative optimization process is performed; if the estimated value does not exceed the preset residual threshold, no further intervention is required, and a leak-free control sequence is finally obtained. In step S11, image sequences, pressure data, and viscosity change data of the fluid during the disconnection process are acquired, including: Real-time data acquisition of the fluid disconnection process is performed using a sensor array to obtain image sequences and pressure data. High-precision sensors are used to monitor minute fluctuations in fluid viscosity in real time and obtain viscosity change data.

[0021] It should be noted that, firstly, the sensor array includes a high-resolution camera and a pressure sensor. The camera is positioned directly above the fluid disconnection area 10 mm below the dispensing needle, with a sampling frequency set to 100 frames per second to ensure complete capture of millisecond-level changes in the necking pattern. The pressure sensor is installed in the dispensing tubing near the needle tip, with a sampling accuracy of once per millisecond, synchronously recording pressure fluctuations when the fluid disconnects. A high-precision viscosity sensor is integrated into the fluid channel of the dispensing valve, with a monitoring accuracy set to 0.1 centipoise, capable of capturing minute fluctuations in fluid viscosity.

[0022] All sensors are equipped with high-precision hardware clocks, which are calibrated every 100 milliseconds via clock synchronization commands from the main control unit to ensure that the timestamp deviation of image sequences, pressure data, and viscosity data is ≤1 millisecond. The acquired raw data is transmitted in real time to the main control unit's circular buffer via industrial Ethernet, with a buffer capacity of 1000 frames to prevent data overflow.

[0023] It is worth noting that the monitoring accuracy is set based on the fact that when the fluid viscosity fluctuates slightly, such as 0.1 centipoise or more, it will significantly affect the disconnection characteristics. Accurate capture is required to ensure the adaptability of subsequent signal waveforms.

[0024] All sensors must undergo full-range calibration before being put into use to ensure measurement accuracy. Camera calibration uses a checkerboard calibration method with a checkerboard size of 10mm × 10mm. Fifteen sets of calibration images are acquired within a distance range of 50mm to 200mm. The intrinsic parameter matrix and distortion coefficients are calculated using the `calibrateCamera` function of the OpenCV library. After calibration, the image distortion rate is controlled within 0.1%, achieving a physical size conversion accuracy of 0.01mm per pixel at a shooting distance of 150mm. Pressure sensor calibration uses a Drucker DPI610 pressure calibrator, calibrating at five calibration points (0, 4, 8, 12, and 16 bar). A linear correction coefficient is fitted, and the measurement error after calibration is ≤0.05 bar. Viscosity sensor calibration uses standard viscosity liquids of 2cm, 5cm, 10cm, and 20cm, performing four-point calibration at a constant temperature of 25℃. A viscosity-torque correction curve is fitted, and the measurement error after calibration is ≤0.05cm.

[0025] To ensure the temporal consistency of multi-dimensional data, a hardware clock calibration scheme is adopted for sensor time synchronization. The main control unit is equipped with a Beidou / GPS dual-mode timing module with a time accuracy of ±10 microseconds. Every 100ms, it sends clock synchronization commands to the camera, pressure sensor, and viscosity sensor. The deviation between the local clock of each sensor and the clock of the main control unit is controlled within 0.1ms to avoid the impact of timestamp deviation on subsequent data fusion and analysis.

[0026] Before data acquisition, bad frames are removed. If the grayscale value of three consecutive frames does not change or the pressure data changes by more than 5 bar, it is identified as a bad frame and removed to avoid affecting subsequent analysis.

[0027] For example, in a pipeline fluid rupture experiment scenario, the sensor array can simultaneously acquire raw records containing 500 frames of image sequences and 5000 sets of pressure data, while the viscosity sensor simultaneously records dynamic data of the fluid viscosity changing from 5.2 centipoise to 3.8 centipoise, with all data timestamp deviations controlled within 0.05 milliseconds.

[0028] In step S12, based on the image sequence and pressure data, the abnormal disconnection time is located, the image corresponding to the abnormal disconnection time is extracted, and the necking diameter and necking length are extracted from it, including: The image sequence is denoised and contrast enhanced, and then a multi-frame boundary point set of the necking region is obtained through edge detection; The multi-frame boundary point set is subjected to temporal tracking and inter-frame matching to extract the parameters of the necking morphology changing over time and determine the dynamic characteristics of the necking morphology. The dynamic features and the pressure data are aligned and fused along the time axis to obtain the correlation features between the neck constriction morphology and pressure changes. If the fluctuation amplitude of the associated feature is higher than the preset amplitude threshold, it is determined to be an abnormal disconnection state and the abnormal disconnection time is located. The image corresponding to the abnormal disconnection time is extracted and enlarged. The necking diameter and necking length are extracted from the enlarged image.

[0029] It should be noted that image preprocessing is divided into two stages: denoising and contrast enhancement. First, median filtering is used for denoising the image sequence, with a filter window size of 3×3. Histogram equalization is used for contrast enhancement, which expands the grayscale range of the necking region from 50-100 to 20-200, improving boundary recognition clarity. Edge detection uses the Canny algorithm, with a low threshold of 80 and a high threshold of 150. This threshold setting is based on the test analysis of 100 sets of fluid necking images; this threshold combination can accurately extract the necking contour and filter background noise. Specifically, the preprocessed image is first Gaussian smoothed, then the gradient magnitude and direction are calculated. Non-maximum suppression is used to remove non-edge pixels. Finally, double threshold detection and edge connection are performed to obtain a multi-frame set of boundary points for the necking region. Each boundary point set contains 50 to 100 pixels, evenly distributed on the contour of the necking region.

[0030] Subsequently, the Lucas-Kanade optical flow method was used for temporal tracking to perform feature matching on the boundary point sets of adjacent frames. This algorithm determines the corresponding position of the boundary point in the previous frame in the next frame by calculating the optical flow vector within a local window. A pyramid of 3 layers was set to improve matching robustness, and 10 iterations were performed to ensure matching accuracy. A 5-pixel threshold for inter-frame matching error was set; when the matching error exceeded this threshold, feature point detection and matching were re-performed to avoid deviations in necking morphology feature extraction due to matching errors. Through temporal tracking, parameters showing the changes in necking width and length over time can be extracted. For example, during the necking formation stage of epoxy resin, the necking width decreases from 5 mm to 1 mm at a rate of 0.02 mm / ms, while the necking length increases from 10 mm to 15 mm, thereby determining the dynamic characteristics of the necking morphology.

[0031] Next, the dynamic features of the neck contraction morphology and pressure data are aligned and fused according to time stamps. The Pearson correlation coefficient is used to calculate the correlation features between the two. The correlation features are the Pearson correlation coefficients of the rate of change of neck contraction width and the rate of change of pressure in the same frame. By calculating the correlation coefficients for consecutive frames, a sequence of correlation features is formed. This sequence can intuitively reflect the degree of coordination between neck contraction morphology and pressure changes. The stronger the coordination, the closer the correlation coefficient is to 1 or -1; the weaker the coordination, the closer the correlation coefficient is to 0.

[0032] The preset threshold for the fluctuation amplitude of the correlation feature is 0.3, and the Pearson correlation coefficient ranges from -1 to 1. This threshold was determined through statistical analysis of disconnection data for different fluids and operating conditions over a period of 6 months. When the fluctuation amplitude of the correlation feature, i.e., the absolute value of the difference between the correlation features of two consecutive frames, exceeds 0.3, it indicates that the synergy between the necking morphology and pressure change has been broken, and the probability of irregular necking during fluid disconnection exceeds 85%, at which point it is judged as an abnormal disconnection state. For example, for epoxy resin adhesive at 2.2 seconds, the correlation feature between the necking width change rate and the pressure change rate in the previous frame was 0.6, and the correlation feature in the current frame suddenly drops to 0.2. The absolute value of the difference between the correlation features between consecutive frames is 0.4, thus the abnormal disconnection time is located at 2.2 seconds.

[0033] When extracting images corresponding to the moment of abnormal disconnection, if the timestamp deviation is ≤10ms (i.e., the acquisition period of one frame), the current frame is directly selected; if the deviation exceeds 10ms, the frame with the closest timestamp is selected and the deviation value is marked to ensure that the extracted image can truly reflect the fluid morphology at the moment of abnormal disconnection. The extracted images are magnified 5 times. Experimental verification shows that this magnification allows clear observation of the minute ripple features caused by surface tension in the necking region without causing image distortion due to excessive magnification. The necking diameter is calculated by fitting the minimum circumcircle to the boundary point set, and the necking length is calculated by fitting the major axis of the rectangle. Combined with the camera calibration results (i.e., 1 pixel corresponds to 0.01mm), the pixel size is converted to physical size. For example, for the image of the abnormal disconnection of epoxy resin adhesive, the diameter of the fitted minimum circumcircle is 0.8 mm, and the major axis of the fitted rectangle is 3 mm, meaning the necking diameter is 0.8 mm and the necking length is 3 mm.

[0034] In step S13, the real-time rate of change of the necking diameter is calculated based on the necking diameter and the necking length. A state indicator indicating that the fluid is about to disconnect is obtained based on the necking length and the real-time rate of change. The timing of the disconnection is predicted to obtain a predicted disconnection timing value, including: The necking diameter and the necking length are standardized to obtain a set of standard parameters; Based on the set of standard parameters, the real-time change rate of the necking diameter is calculated. If the necking diameter is less than a preset necking diameter threshold and the real-time change rate shows an increasing trend, the fluid state is determined to be about to disconnect, and the corresponding state identifier is obtained. Based on the status identifier, a time series analysis is performed on the standard parameter set to obtain the predicted disconnection timing value.

[0035] It should be noted that, firstly, standardization is achieved through data cleaning and Min-Max normalization. Data cleaning uses the 3σ criterion to remove outliers. First, the mean μ and standard deviation σ of the original data for the necking diameter and length are calculated. Data exceeding the range of μ ± 3σ are identified as outliers and removed to avoid extreme data interfering with subsequent calculations.

[0036] In the Min-Max normalization stage, different dimensional parameters of necking diameter and length are mapped to the range of 0 to 1. The diameter is normalized to a range of 0-5 mm, and the length is normalized to a range of 0-10 mm. This range is derived from actual data statistics of precision dispensing scenarios such as electronic packaging and medical device manufacturing within 6 months. The statistics show that more than 98% of the fluid necking diameter is concentrated in 0-5 mm, and the necking length is concentrated in 0-10 mm, covering the necking characteristic range of mainstream dispensing fluids.

[0037] If the original data contains special values ​​that exceed this range, a truncation process is adopted: if the diameter is greater than 5 mm, take 5 mm; if the length is greater than 10 mm, take 10 mm. Then, it is included in the normalization calculation to avoid extreme values ​​from causing distortion of the normalization results, ensure that the dimensional differences are effectively eliminated, and not affect the accuracy of the subsequent calculation of the real-time change rate of the necking diameter.

[0038] Subsequently, the real-time change rate of the necking diameter was calculated using the sliding window method, with a window size of 5 frames, corresponding to 50 milliseconds. This window size was experimentally verified to ensure real-time performance while smoothing fluctuations across multiple frames of data, avoiding misjudgments caused by errors in a single frame. The calculation formula is as follows: Where r is the real-time rate of change of the necking diameter, and D n D is the necking diameter of the current frame. n-4 The diameter of the neck contraction of the first frame within the window. The time interval corresponding to the window is 50 milliseconds. The preset necking diameter threshold is 0.1 mm, which has been statistically analyzed through numerous experiments. When the diameter is less than 0.1 mm and the absolute value of the rate of change shows an increasing trend, the probability of the fluid disconnecting within 0.5 seconds exceeds 90%. If the above conditions are met, the fluid state is determined to be about to disconnect, and the status flag is set to the number 1, with the flag including the current necking parameters.

[0039] Simultaneously, necking length data should be used to assist in verifying the impending breakage state, avoiding misjudgments caused by relying solely on necking diameter and rate of change. This auxiliary verification logic for necking length is based on experimental statistics. In 500 breakage experiments with different fluids and operating conditions, over 90% of the fluids about to break showed a stable necking length within the range of 2-5 mm. Moreover, when the necking diameter is less than 0.1 mm and the absolute value of the rate of change shows an increasing trend, the necking length exhibits a characteristic of first slightly increasing and then stabilizing. This is because before the fluid breaks, under the action of surface tension and shear force, the length of the necking region will first stretch to a critical value before entering a morphologically stable stage before fracture.

[0040] The specific verification method is as follows: the preset critical range for necking length is 2-5 mm. When the necking diameter is less than 1.0 mm and the absolute value of the rate of change shows an increasing trend, if the necking length is simultaneously within 2-5 mm and there is no significant increase for three consecutive frames, with an increase of ≤0.1 mm / frame, the fluid state is further confirmed as about to break, and the status indicator is set to the number 1. If the necking length is not within 2-5 mm, for example, only 0.8 mm or reaching 6.2 mm, even if the diameter and rate of change meet the conditions, it is not temporarily determined as about to break, and the next frame of data needs to be re-monitored to avoid misjudgment caused by abnormal necking. Abnormal necking includes short necking caused by local fluid adhesion and long necking caused by excessive stretching.

[0041] For example, in the detection of epoxy resin adhesive, the necking diameter is 0.09 mm, which is less than 0.1 mm, and the rate of change changes from -0.02 mm / ms to -0.05 mm / ms, meaning the absolute value increases. Simultaneously, the necking length of 3.2 mm falls within the 2-5 mm range, and the increase over three consecutive frames is 0.05 mm / frame ≤ 0.1 mm / frame. Using the necking length as an auxiliary verification method, the impending disconnection state can be more accurately determined. If the necking length is only 0.6 mm at this point, below the critical range, it is determined to be a short necking caused by local abnormal adhesion, and the impending disconnection indicator is not triggered; further monitoring of subsequent necking morphology changes is necessary.

[0042] When performing time-series analysis on the standard parameter set based on status flags, key data valuable for predicting disconnection timing is selected through status flag filtering, avoiding indiscriminate analysis of all frame data that could lead to a decrease in prediction accuracy. Status flags are primarily used to filter effective analysis phases; the ARIMA model's time-series analysis process is only initiated when the status flag is 1; if the status flag is 0, time-series analysis is temporarily suspended to reduce unnecessary computation.

[0043] The ARIMA model was adopted, with parameters p=2, d=1, and q=1, where p is the number of autoregressive terms, d is the difference order, and q is the number of moving average terms. This parameter combination was determined after multiple adjustments and achieved the highest fitting accuracy for the necking parameter time series. Inputting a standardized necking parameter time series of nearly 20 frames, the model analyzes the parameter variation trends and outputs predicted values ​​for the disconnection timing.

[0044] For example, after standardization, the normalized value of the necking diameter decreased from 0.22 to 0.18, and the normalized value of the real-time change rate of the necking diameter changed from -0.004 to -0.01. Combined with the normalized value of the necking length stabilizing at 0.32, it was determined that the necking was about to break. Inputting these standardized parameters into the ARIMA model, the model predicted the breakage time to be 0.4 seconds later.

[0045] In step S14, the predicted disconnection timing value and the viscosity change data are integrated to obtain a comprehensive dataset. Core parameters related to signal amplitude and timing are extracted from the comprehensive dataset and calibrated. Based on the calibrated parameters, a signal waveform adapted to the fluid disconnection characteristics is generated, including: By integrating the predicted disconnection timing with the viscosity change data, a comprehensive dataset is obtained; Dynamic simulation analysis is performed on the comprehensive dataset to extract core parameters related to signal amplitude and timing, thereby obtaining dynamic characteristic values; wherein, the dynamic simulation analysis is performed using a pre-established fluid database matching method; If the dynamic characteristic value exceeds the preset dynamic characteristic value threshold, the signal amplitude and timing are calibrated to obtain a combination of calibration parameters; Based on the combination of calibration parameters, a signal waveform adapted to the fluid disconnection characteristics is generated.

[0046] It should be noted that, firstly, when integrating the predicted disconnection timing with viscosity change data, the predicted disconnection timing is matched with the corresponding viscosity data, and the necking parameters are supplemented to form a comprehensive dataset containing three-dimensional information of time, morphology, and physical properties.

[0047] For example, taking the monitoring of the disconnection process of epoxy resin adhesive as an example, the time dimension of its comprehensive dataset includes two core data points: first, the current monitoring timestamp t=2.2 seconds, which is aligned with the timestamps of the image sequence and pressure data collected by the sensor; second, the predicted disconnection timing value 0.4 seconds later, that is, the predicted fluid will disconnect at t=2.6 seconds. The morphological dimension consists of standardized necking parameters, including a normalized value of 0.18 for the necking diameter, a normalized value of 0.32 for the necking length, and a normalized value of -0.01 for the real-time change rate of the necking diameter. The physical properties dimension includes viscosity change data, including the current viscosity value of 3.8 centipoise and the viscosity change of -0.2 centipoise within the last 50 milliseconds, i.e., from 4.0 centipoise to 3.8 centipoise. The above data together constitute a complete comprehensive dataset record, which can be expressed in the format of "timestamp: 2.2 seconds, disconnection timing prediction: 0.4 seconds later; normalized value of necking diameter: 0.18, normalized value of necking length: 0.32, normalized value of necking diameter change rate: -0.01; current viscosity: 3.8 centipoise, viscosity change in 50 ms: -0.2 centipoise", providing complete multi-dimensional data support for subsequent dynamic simulation analysis.

[0048] Next, dynamic simulation analysis was performed on the comprehensive dataset. Considering the response speed requirements of real-time control scenarios, a lightweight solution of pre-simulation modeling + real-time parameter matching was adopted to replace the full-process real-time simulation, avoiding control delays caused by excessive computation. Specifically, before system deployment, pre-simulation was performed in advance using the fluid flow module of COMSOL Multiphysics software for 10 mainstream dispensing fluids commonly used in electronic packaging and medical device manufacturing scenarios, such as silicone, epoxy resin, and UV adhesives, as well as typical working conditions of 1-5 bar pressure and 1-5 m / s flow rate. For example, for water-based fluids, the viscosity was set to 3 cP, the density to 1000 kg / m³, and the boundary condition to be no slip for simulation. Dynamic characteristic values ​​such as signal amplitude peak and timing period were extracted from the simulation results and stored locally in the main control unit.

[0049] During real-time control, there is no need to re-execute the complete simulation. Instead, key parameters of the current comprehensive dataset, such as fluid type, current viscosity, normalized necking parameters, and predicted disconnection timing, are quickly matched against a pre-established database. If a completely identical parameter combination exists, the corresponding dynamic characteristic value is directly called. If there is a slight deviation in the parameters, linear interpolation is used to calculate the corrected dynamic characteristic value. The entire matching and correction process takes ≤10 milliseconds, meeting the requirements of real-time control. For example, in real-time monitoring of epoxy resin, after the comprehensive dataset shows that the fluid type is epoxy resin, the current viscosity is 3.8 centipoise, the normalized necking diameter is 0.18, and the predicted disconnection timing is 0.4 seconds, the database matches the record "epoxy resin, viscosity 4.0 centipoise, normalized necking diameter 0.2, dynamic characteristic value 3.6 units, 0.26 seconds". After linear interpolation correction, the current operating condition's dynamic characteristic value is 3.5 units, 0.25 seconds, with an error of ≤5% compared to the full-process simulation result. This significantly reduces the computational load in the real-time control stage while ensuring calculation accuracy.

[0050] It is worth noting that in industrial dispensing scenarios, when the signal amplitude exceeds 3.0 units, the control signal is prone to exceed the response range of the actuator, causing the fluid disconnection control to fail. Therefore, the amplitude threshold of the preset power characteristic value is set to 3.0 units.

[0051] Furthermore, the "unit" here refers to the standardized expression of the actuator drive voltage, which is essentially tied to the output range of the hardware drive circuit. This solution uses the DAC module of the STM32F407 microcontroller to output a 0-3.3V analog signal, which is amplified to 0-10V by a power amplifier to drive the actuator. To adapt to the drive voltage range of different actuator models, the amplified maximum drive voltage of 10V is assigned to 5 units, thus determining that 1 unit corresponds to a 2V drive voltage. The "unit" mentioned thereafter will follow this standardized definition.

[0052] If the dynamic characteristic value exceeds the threshold, PID control is used for calibration with a proportional coefficient of 0.8, an integral coefficient of 0.2, and a derivative coefficient of 0.1. This parameter combination, after debugging, can quickly eliminate amplitude deviation without overshoot. The signal amplitude is adjusted according to the amplitude deviation, and the timing is calibrated according to the rule of shortening the timing period by 0.05 seconds for every 1 centipoise increase in viscosity, thus obtaining the calibration parameter combination.

[0053] A square wave signal waveform is generated based on the calibration parameter combination. An STM32F407 microcontroller's DAC module, configured with a 1MHz sampling rate, outputs a 0-3.3V analog signal, which is amplified to 0-10V by a power amplifier, corresponding to a signal amplitude of 0-5 units. The duty cycle of the square wave is set to 50%, which has been experimentally verified to ensure the intervention effect while preventing overheating of the actuator due to prolonged high-load operation. The generated signal waveform is observed and verified using an oscilloscope, ensuring that the amplitude is stable at 3.1 units ± 0.1 units and the period is stable at 0.16 seconds ± 0.01 seconds, consistent with the calibration parameters and adapted to the fluid disconnection characteristics.

[0054] For example, in the scenario of controlling the disconnection of epoxy resin adhesive, after dynamic simulation analysis and parameter calibration, the determined calibration parameter combination is a signal amplitude of 3.1 units and a period of 0.16 seconds. When generating a square wave signal according to the above hardware configuration, the STM32F407 microcontroller DAC module outputs a 2.013V analog signal, which is amplified by a power amplifier to output a 6.2V drive voltage. Oscilloscope observation shows that this voltage is stable at 6.2V ± 0.2V, and the square wave period is stable at 0.16 seconds ± 0.01 seconds, which perfectly matches the calibration parameters. When this signal waveform is applied to the epoxy resin adhesive dispensing process, the fluid necking area shrinks smoothly from 0.9 mm to 0.3 mm under the action of the signal before disconnecting, without dripping or residue, verifying the adaptability of the signal waveform to the fluid disconnection characteristics.

[0055] In step S15, real-time flow velocity data is acquired, the signal waveform is specifically corrected based on the real-time flow velocity data, and an optimized control signal is generated based on the corrected signal waveform, including: Real-time flow velocity data is collected by a flow sensor and combined with a preset flow velocity threshold to determine whether the flow velocity exceeds the limit. If it does not exceed the limit, the signal waveform is used as the correction waveform; if it exceeds the limit, it is marked as the waveform to be corrected. Based on the degree of deviation of the real-time flow rate data from the preset flow rate threshold, the amplitude and timing parameters of the waveform to be corrected are specifically modified to obtain the corrected waveform. Based on the corrected waveform, an optimized control signal is generated.

[0056] It should be noted that real-time flow rate data is acquired via a flow sensor, specifically the Hydac HFS200 model, with a measurement range of 0-10 m / s, an accuracy of ±0.5%FS, and a response time ≤10ms. Installed at the dispensing nozzle, it directly acquires the actual dispensing velocity, avoiding measurement errors caused by pipeline friction. The sensor's output pulse signal is converted into a digital signal by an Advantech ADAM-4050 counter module, with a sampling frequency set to 100Hz, meaning the flow rate data is updated every 10ms to ensure real-time capture of flow rate fluctuations.

[0057] The preset flow rate threshold is 3.0 m / s. This threshold is a commonly used flow rate range in precision dispensing scenarios such as electronic packaging and medical device manufacturing. Exceeding this range will significantly alter the fluid flow state. Real-time flow rate data is compared to the threshold. If the flow rate is within the threshold, the signal waveform is directly used as the correction waveform; if the flow rate exceeds the threshold, the waveform is corrected according to the degree of deviation. For every 0.1 m / s increase in flow rate, the signal amplitude is reduced by 0.1 units, and the timing response time is shortened by 0.02 seconds. This rule has been verified through 100 waveform correction experiments at different flow rates, effectively suppressing turbulence caused by excessively high flow rates, restoring the fluid to a laminar flow state, and ensuring that the corrected waveform matches the real-time flow rate.

[0058] The corrected waveform parameters, such as amplitude of 2.3 units and timing response time of 10 milliseconds, are converted into a 0-10V analog voltage signal and transmitted to the actuator drive unit via a CAN bus at a transmission rate of 1 Mbps. The frame structure uses standard data frames to ensure real-time performance and reliability. Upon receiving the signal, the drive unit activates the piezoelectric ceramic actuator within 10 milliseconds, applying a control signal of the corresponding amplitude. For example, after the optimized control signal transmission for epoxy resin adhesive, the flow velocity drops from 3.8 m / s to 2.8 m / s, returning to the threshold range, the fluid morphology stabilizes again, and the necking rate tends to level off.

[0059] In step S16, key intervention points are extracted from the optimized control signal, pulse intervention is applied according to the triggering timing corresponding to the key intervention points, and then the estimated value of the colloid residue is calculated, including: Extract key intervention points that are time-series correlated with the peak value of the necking characteristic from the optimized control signal. If the amplitude of the key intervention point does not meet the preset amplitude threshold, then re-extract the key intervention point and verify it. Obtain the trigger timing corresponding to the key intervention point, apply pulse interference according to the trigger timing, and then calculate the estimated value of the colloid residue.

[0060] It should be noted that, firstly, the key intervention point extraction adopts a sliding window peak detection algorithm, with the window size set to 3 signal cycles. This window size can cover the complete signal fluctuation cycle, ensuring that no peaks are missed. The algorithm first calculates the maximum signal amplitude within the window. If this maximum value exceeds the preset amplitude threshold for intervention points, it is determined to be a key intervention point; if it does not exceed the threshold, the window is moved to continue detection.

[0061] It is worth noting that the threshold for intervention point amplitude is set based on the effective amplitude test of pulse intervention. When the amplitude is less than 4.0 units, the evolution trend of neck constriction characteristics cannot be changed.

[0062] The trigger timing is set to 0.1 seconds before the necking characteristic reaches its peak. The peak of the necking characteristic is determined jointly by a pressure sensor and an image sequence. When the pressure data reaches its maximum value and the necking diameter in the image reaches its minimum value, the peak moment of the necking characteristic is determined. For example, the peak moment of the necking characteristic of epoxy resin adhesive is 2.6 seconds, and the trigger timing is set to 2.5 seconds. This advance amount has been verified by fluid dynamics experiments and can intervene at the most sensitive stage of the necking characteristic, suppressing leakage to the greatest extent.

[0063] The estimation of colloidal residue was performed using a visual inspection method. A camera was used to capture images of the fluid residue area after intervention, and the area of ​​the residue area was extracted. The estimated value was calculated based on the calibration relationship that 1 square millimeter corresponds to 0.001 microliters. Simultaneously, the estimation was verified by weighing with an electronic balance. An error ≤5% was considered valid. For example, an intervention point with an amplitude of 4.5 units was extracted, and the trigger timing was set to 0.1 seconds before the necking peak. After applying a 24V pulse intervention, the visual inspection showed a residue area of ​​0.3 square millimeters, and the calculated residue amount was 0.0003 microliters. The weighing verification showed an error of 3%, indicating the estimated value was valid.

[0064] In step S17, if the estimated value exceeds a preset residual threshold, an iterative optimization process is performed; if the estimated value does not exceed the preset residual threshold, no further intervention is required, ultimately resulting in a leak-free control sequence. The iterative optimization process includes: The records and results of this intervention are obtained and input into a preset neural network model. A set of optimized intervention parameters are output and the intervention is performed again. The updated estimate of the amount of colloid residue after the intervention is calculated. The process of obtaining the records and results of this intervention and inputting them into a preset neural network model is repeated, outputting a set of optimized intervention parameter combinations and performing the intervention again, and calculating the updated estimate of the colloid residue after the intervention, continues until the updated estimate is lower than the preset residue threshold.

[0065] It should be noted that, firstly, the preset residual threshold is 0.0005 microliters. This threshold was determined through testing in two core precision manufacturing scenarios: electronic packaging, such as chip bonding adhesive dispensing, and medical device manufacturing, such as catheter bonding adhesive dispensing.

[0066] The iteratively optimized neural network model has a three-layer structure: an input layer, a hidden layer, and an output layer. The input layer contains eight neurons, corresponding to the current intervention amplitude, current intervention period, current trigger timing, current residual quantity estimate, current fluid viscosity, real-time flow rate, normalized necking diameter, and normalized necking length, respectively. The current intervention amplitude refers to the signal amplitude of the previous pulse intervention, in standardized units; the current intervention period refers to the square wave period of the previous intervention signal, in seconds; the current trigger timing refers to the lead time of the previous intervention signal relative to the peak of the necking characteristic, in milliseconds; the current residual quantity estimate refers to the colloidal residual quantity calculated after the previous intervention, in microliters; the current fluid viscosity is the real-time monitored fluid viscosity value, in centipoises; the real-time flow rate is the actual outflow velocity of the fluid during intervention, in meters per second; and the normalized necking diameter and necking length values ​​are the standardized results of the necking diameter and length during intervention, respectively, both ranging from 0 to 1. These eight input items are all core parameters that affect the effectiveness of the intervention and can comprehensively reflect the current working conditions and feedback from the previous round of intervention.

[0067] The hidden layer contains 16 neurons and uses the ReLU activation function, which can effectively alleviate the gradient vanishing problem and is suitable for the nonlinear correlation between parameters and residual quantities in industrial scenarios.

[0068] The output layer contains three neurons, corresponding to the optimized intervention amplitude, optimized intervention period, and optimized trigger timing, respectively. The optimized intervention amplitude is the adjusted standardized unit, the optimized intervention period is the adjusted square wave period in seconds, and the optimized trigger timing is the adjusted advance time in milliseconds. These three directly output the key parameters for the next round of intervention.

[0069] The model training data comes from 1000 sets of intervention records for different fluids and operating conditions. 800 sets serve as the training set, covering 8 fluids and 40 operating condition combinations, while 200 sets serve as the validation set, covering the remaining 2 fluids and 10 operating condition combinations. The training process uses mean squared error as the loss function and iterates for 500 rounds. After each round, the fitting effect is verified using the validation set. The final model's prediction accuracy on the validation set is ≥95% of the actual intervention effect (i.e., residual error). In other words, in over 95% of the validation samples, the residual error corresponding to the optimized parameters output by the model is ≤5%, demonstrating stable output of intervention parameters adapted to the operating conditions.

[0070] Each iteration inputs the intervention record into the model, outputs optimized parameters, executes the intervention again, and calculates and updates the residual value. The iteration terminates when the residual value falls below a threshold or the number of iterations reaches 10, to avoid infinite iterations due to slow model convergence. If the residual value still does not fall below the threshold after 10 iterations, the current optimal parameters are output, and an alarm is triggered, prompting manual inspection of the equipment status.

[0071] Ultimately, the drip-free control sequence is a parameter array that includes pulse amplitude, duration, trigger timing, and flow rate correction coefficient. Different fluid types are adapted according to rules. For every 1 centipoise increase in viscosity, the necking diameter threshold is lowered by 0.1 mm to ensure compatibility with different fluids such as silicone and epoxy resin.

[0072] For example, if the initial residual amount is 0.0008 microliters, which exceeds the threshold, the output after inputting into the model is an optimization amplitude of 4.8 units, a period of 0.03 seconds, and a trigger timing of 0.08 seconds in advance. After another intervention, the residual amount is 0.0002 microliters, which is below the threshold, and the iteration stops. The control sequence is determined to be an amplitude of 4.8 units, a period of 0.03 seconds, a trigger timing of 0.08 seconds in advance, and a flow rate correction coefficient of 0.9.

[0073] In summary, this invention discloses an anti-drip adaptive fluid dispensing control method, comprising acquiring image sequences, pressure data, and viscosity change data of the fluid during the disconnection process. Based on the image sequences and pressure data, the abnormal disconnection moment is located, and the corresponding image is extracted to obtain the necking diameter and necking length. The real-time rate of change is calculated based on the necking parameters to determine the impending disconnection state and predict the disconnection timing. The predicted disconnection timing value and viscosity data are integrated, and a suitable signal waveform is generated after kinetic simulation calibration. The waveform is corrected by combining real-time flow rate to generate an optimized control signal. Key intervention points are extracted, pulse intervention is applied, and the residual amount of colloid is calculated. If the residual amount exceeds a preset threshold, the parameters are iteratively optimized through a neural network, ultimately obtaining a drip-free control sequence.

[0074] This invention uses multi-dimensional sensors to collaboratively collect fluid morphology, mechanical and physical property data, relies on image processing and dynamic simulation to accurately analyze the necking characteristics, dynamically correlates flow rate and viscosity adjustment control signals, and continuously corrects intervention parameters with an iterative optimization mechanism. This achieves drip-free control of the fluid dispensing process, significantly improving control accuracy and stability. It can adapt to fluid conditions with different viscosities and flow rates, meeting the high-precision requirements of fluid dispensing in precision manufacturing scenarios such as electronic packaging and medical device manufacturing.

[0075] Reference Figure 2 The second embodiment of the present invention provides an anti-drip adaptive fluid dispensing control system, comprising: The data acquisition module is used to acquire image sequences, pressure data, and viscosity change data of the fluid during the disconnection process; The feature extraction module is used to locate the abnormal disconnection time based on the image sequence and pressure data, extract the image corresponding to the abnormal disconnection time, and extract the necking diameter and necking length from it; The disconnection prediction module is used to calculate the real-time change rate of the necking diameter based on the necking diameter and the necking length, obtain the state indicator that the fluid is about to disconnect based on the necking length and the real-time change rate, and predict the disconnection timing to obtain the disconnection timing prediction value. The waveform generation module is used to integrate the predicted disconnection timing value and the viscosity change data to obtain a comprehensive dataset. It extracts and calibrates the core parameters related to signal amplitude and timing from the comprehensive dataset and generates a signal waveform adapted to the fluid disconnection characteristics based on the calibrated parameters. The feedback optimization module is used to acquire real-time flow velocity data, perform targeted correction on the signal waveform based on the real-time flow velocity data, and generate an optimized control signal based on the corrected signal waveform. The pulse intervention module is used to extract key intervention points in the optimized control signal, apply pulse intervention according to the triggering sequence corresponding to the key intervention points, and then calculate the estimated value of the colloidal residue. The iterative calibration module is used to perform an iterative optimization process if the estimated value exceeds the preset residual threshold, and if the estimated value does not exceed the preset residual threshold, no further intervention is required, ultimately obtaining a leak-free control sequence.

[0076] It should be noted that the anti-drip adaptive fluid dispensing control system provided in this embodiment of the invention is used to execute all the process steps of the anti-drip adaptive fluid dispensing control method of the above embodiment. The working principle and beneficial effect of the two are one-to-one, so they will not be described again.

[0077] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a data acquisition program. When the processor executes the computer program, it implements the steps in the various embodiments of the anti-drip adaptive fluid dispensing control method described above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above system embodiments, such as the feedback optimization module.

[0078] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0079] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0080] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0081] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0082] If the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or system capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0083] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0084] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A drip-proof adaptive fluid dispensing control method, characterized in that, include: Acquire image sequences, pressure data, and viscosity change data of the fluid during the disconnection process; Based on the image sequence and pressure data, the abnormal disconnection time is located, and the image corresponding to the abnormal disconnection time is extracted, from which the necking diameter and necking length are extracted. This includes: denoising and contrast enhancement of the image sequence, and then obtaining a multi-frame boundary point set of the necking region through edge detection; performing temporal tracking and inter-frame matching on the multi-frame boundary point set to extract the necking morphology change parameters over time and determine the dynamic features of the necking morphology; aligning and fusing the dynamic features and the pressure data along the time axis to obtain the correlation features between the necking morphology and pressure changes; if the fluctuation amplitude of the correlation features is higher than a preset amplitude threshold, it is determined to be an abnormal disconnection state and the abnormal disconnection time is located, the image corresponding to the abnormal disconnection time is extracted and enlarged, and the necking diameter and necking length are extracted from the enlarged image. The real-time rate of change of the necking diameter is calculated based on the necking diameter and the necking length. The state indicator that the fluid is about to be disconnected is obtained according to the necking length and the real-time rate of change. The timing of disconnection is predicted to obtain the predicted value of disconnection timing. Integrating the predicted disconnection timing with the viscosity change data yields a comprehensive dataset. Core parameters related to signal amplitude and timing are extracted from this dataset and calibrated. Based on the calibrated parameters, a signal waveform adapted to the fluid disconnection characteristics is generated. This process includes: performing dynamic simulation analysis on the comprehensive dataset to extract core parameters related to signal amplitude and timing, obtaining dynamic characteristic values; wherein the dynamic characteristic values ​​include the signal amplitude peak and timing period, and the dynamic simulation analysis uses a pre-established fluid database matching method; if the dynamic characteristic values ​​exceed a preset dynamic characteristic value threshold, the signal amplitude and timing are calibrated to obtain a calibration parameter combination; and based on the calibration parameter combination, a signal waveform adapted to the fluid disconnection characteristics is generated. Acquire real-time flow velocity data, and based on the degree of deviation of the real-time flow velocity data from a preset flow velocity threshold, perform targeted correction on the amplitude and timing parameters of the signal waveform, and generate an optimized control signal based on the corrected signal waveform; Key intervention points that are time-series correlated with the peak value of the neck contraction characteristic are extracted from the optimized control signal. The peak value of the neck contraction characteristic is the moment when the pressure data reaches its maximum value and the neck contraction diameter in the image reaches its minimum value. The key intervention point extraction adopts a sliding window peak detection algorithm. First, the maximum value of the signal amplitude within the window is calculated. If the maximum value exceeds the preset amplitude threshold of the intervention point, it is determined to be a key intervention point. Pulse intervention is applied according to the triggering sequence corresponding to the key intervention point, and then the estimated value of the colloid residue is calculated. If the estimated value exceeds the preset residue threshold, an iterative optimization process is performed. If the estimated value does not exceed the preset residue threshold, no further intervention is required, and finally a drip-free control sequence is obtained.

2. The anti-drip adaptive fluid dispensing control method according to claim 1, characterized in that, The acquisition of image sequences, pressure data, and viscosity change data of the fluid during the disconnection process includes: Real-time data acquisition of the fluid disconnection process is performed using a sensor array to obtain image sequences and pressure data. High-precision sensors are used to monitor minute fluctuations in fluid viscosity in real time and obtain viscosity change data.

3. The anti-drip adaptive fluid dispensing control method according to claim 1, characterized in that, The process of calculating the real-time rate of change of the necking diameter based on the necking diameter and the necking length, obtaining a state indicator indicating that the fluid is about to disconnect based on the necking length and the real-time rate of change, and predicting the disconnection timing to obtain a predicted disconnection timing value includes: The necking diameter and the necking length are standardized to obtain a set of standard parameters; Based on the set of standard parameters, the real-time change rate of the necking diameter is calculated. If the necking diameter is less than a preset necking diameter threshold and the real-time change rate shows an increasing trend, the fluid state is determined to be about to disconnect, and the corresponding state identifier is obtained. Based on the status identifier, a time series analysis is performed on the standard parameter set to obtain the predicted disconnection timing value.

4. The anti-drip adaptive fluid dispensing control method according to claim 1, characterized in that, The process of acquiring real-time flow velocity data, and based on the degree of deviation of the real-time flow velocity data from a preset flow velocity threshold, specifically correcting the amplitude and timing parameters of the signal waveform, and generating an optimized control signal based on the corrected signal waveform, includes: Real-time flow velocity data is collected by a flow sensor and combined with a preset flow velocity threshold to determine whether the flow velocity exceeds the limit. If it does not exceed the limit, the signal waveform is used as the correction waveform; if it exceeds the limit, it is marked as the waveform to be corrected. Based on the degree of deviation of the real-time flow rate data from the preset flow rate threshold, the amplitude and timing parameters of the waveform to be corrected are specifically modified to obtain the corrected waveform. Based on the corrected waveform, an optimized control signal is generated.

5. The anti-drip adaptive fluid dispensing control method according to claim 1, characterized in that, The process involves extracting key intervention points from the optimized control signal that are correlated with the peak timing of the necking characteristic, inputting these key intervention points into the actuator system, applying pulse intervention when the necking characteristic reaches its peak according to the trigger timing corresponding to the key intervention points, and calculating an estimated value of the colloidal residue. Extract key intervention points that are time-series correlated with the peak value of the necking characteristic from the optimized control signal. If the amplitude of the key intervention point does not meet the preset amplitude threshold, then re-extract the key intervention point and verify it. Obtain the trigger timing corresponding to the key intervention point, apply pulse intervention according to the trigger timing, and then calculate the estimated value of the colloidal residue.

6. The anti-drip adaptive fluid dispensing control method according to claim 1, characterized in that, The iterative optimization process includes: The records and results of this intervention are obtained and input into a preset neural network model. A set of optimized intervention parameters are output and the intervention is performed again. The updated estimate of the amount of colloid residue after the intervention is calculated. The process of obtaining the records and results of this intervention and inputting them into a preset neural network model is repeated, outputting a set of optimized intervention parameter combinations and performing the intervention again, and calculating the updated estimate of the colloid residue after the intervention, continues until the updated estimate is lower than the preset residue threshold.

7. A drip-proof adaptive fluid dispensing control system, characterized in that, include: The data acquisition module is used to acquire image sequences, pressure data, and viscosity change data of the fluid during the disconnection process; The feature extraction module is used to locate the abnormal disconnection time based on the image sequence and pressure data, extract the image corresponding to the abnormal disconnection time, and extract the necking diameter and necking length from it. This includes: denoising and contrast enhancement of the image sequence, then obtaining a multi-frame boundary point set of the necking region through edge detection; performing temporal tracking and inter-frame matching on the multi-frame boundary point set to extract parameters of the necking morphology changing over time and determine the dynamic features of the necking morphology; aligning and fusing the dynamic features and the pressure data along the time axis to obtain the correlation features between the necking morphology and pressure changes; if the fluctuation amplitude of the correlation features is higher than a preset amplitude threshold, it is determined to be an abnormal disconnection state and the abnormal disconnection time is located; the image corresponding to the abnormal disconnection time is extracted and enlarged; and the necking diameter and necking length are extracted from the enlarged image. The disconnection prediction module is used to calculate the real-time change rate of the necking diameter based on the necking diameter and the necking length, obtain the state indicator that the fluid is about to disconnect based on the necking length and the real-time change rate, and predict the disconnection timing to obtain the disconnection timing prediction value. The waveform generation module integrates the predicted disconnection timing value with the viscosity change data to obtain a comprehensive dataset. It extracts and calibrates core parameters related to signal amplitude and timing from the comprehensive dataset, and generates a signal waveform adapted to the fluid disconnection characteristics based on the calibrated parameters. This includes: performing dynamic simulation analysis on the comprehensive dataset to extract core parameters related to signal amplitude and timing, obtaining dynamic characteristic values; wherein the dynamic characteristic values ​​include the signal amplitude peak value and timing period, and the dynamic simulation analysis is performed using a pre-established fluid database matching method; if the dynamic characteristic values ​​exceed a preset dynamic characteristic value threshold, the signal amplitude and timing are calibrated to obtain a calibration parameter combination; and a signal waveform adapted to the fluid disconnection characteristics is generated based on the calibration parameter combination. The feedback optimization module is used to acquire real-time flow velocity data, and based on the real-time flow velocity data, to specifically correct the amplitude and timing parameters of the signal waveform for deviations exceeding a preset flow velocity threshold, and to generate an optimized control signal based on the corrected signal waveform. The pulse intervention module is used to extract key intervention points in the optimized control signal that are time-series related to the peak value of the neck contraction characteristic. The peak value of the neck contraction characteristic is the moment when the pressure data reaches its maximum value and the neck contraction diameter in the image reaches its minimum value. The key intervention point extraction adopts a sliding window peak detection algorithm. First, the maximum value of the signal amplitude within the window is calculated. If the maximum value exceeds the preset amplitude threshold of the intervention point, it is determined to be a key intervention point. Pulse intervention is applied according to the triggering sequence corresponding to the key intervention point, and then the estimated value of the colloidal residue is calculated; The iterative calibration module is used to perform an iterative optimization process if the estimated value exceeds the preset residual threshold, and if the estimated value does not exceed the preset residual threshold, no further intervention is required, ultimately obtaining a leak-free control sequence.

Images