CN-121971889-A - Self-adaptive vacuum defoaming process optimization method based on intelligent control system

Abstract

The invention provides a self-adaptive vacuum defoaming process optimization method based on an intelligent control system, which aims at solving the problems that the traditional defoaming technology cannot adapt to microbubble residues and excessively high energy consumption caused by different viscosity liquids, and realizes accurate defoaming through a multidimensional sensing technology and strategy generation and regulation. The method comprises the steps of presetting high-precision vacuum parameters and viscosity grading strategies, establishing a liquid characteristic vector model by adopting YOLOv and Hough transformation algorithms, combining weighted cosine similarity to match a historical optimal strategy, generating a defoaming strategy based on a pre-training network and a multi-target optimization algorithm, verifying the defoaming strategy in a virtual environment, implementing the defoaming in four stages, realizing efficient foam breaking through sectional PID control, resonance vibration and cooperation of a two-stage pump, integrating pressure, temperature and vibration data by a Kalman filtering algorithm, adjusting vacuum pump power and heating strategies, constructing a multi-index evaluation system, and performing iterative optimization through a reward signal driving strategy.

Inventors

• LU LEI
• SHI MINGJUN

Assignees

• 湖北联新显示科技有限公司

Dates

Publication Date

20260505

Application Date

20251226

Claims (7)

1. 1. An adaptive vacuum defoaming process optimization method based on an intelligent control system is characterized by comprising the following steps of: s1, performing equipment inspection and parameter calibration on a defoaming operation procedure, and setting initial operation parameters according to the characteristics of liquid to be treated; S2, detecting liquid characteristics through a multidimensional sensing technology, matching with a preset characteristic database, and constructing a liquid characteristic vector model; S3, loading a pre-trained network model, generating a defoaming strategy by adopting a multi-objective optimization algorithm in combination with defoaming efficiency, product quality and energy consumption indexes, and verifying the defoaming strategy in a virtual environment; S4, executing a defoaming strategy comprising four continuous process stages, wherein the defoaming strategy comprises a mild pre-defoaming treatment stage, an enhanced defoaming treatment stage, a fine defoaming treatment stage and a stabilizing treatment stage in sequence; s5, fusing multisource monitoring data based on a Kalman filtering algorithm, analyzing working condition states, adjusting parameters of a defoaming process, establishing a pressure-temperature safety control mechanism, and configuring an automatic abnormal condition processing scheme; S6, constructing a multi-index quality evaluation system to evaluate the defoaming effect, detecting the bubble residual rate, counting product consistency data, storing the defoaming process data, analyzing the relevance of the technological parameters and the defoaming effect, and updating a decision strategy based on a reward signal generated by the defoaming effect.
2. 2. The method for optimizing an adaptive vacuum defoaming process based on an intelligent control system according to claim 1, wherein the step S1 further comprises: Performing system calibration on the vacuum pump, the pressure gauge and the pressure regulating valve, performing zero calibration and sensitivity verification on the temperature sensor, the viscosity sensor and the bubble detection sensor, running a control self-diagnosis program, and performing tightness test on the vacuum container and the pipeline; The method comprises the steps of presetting process control parameters, wherein the process control parameters comprise a maximum vacuum degree threshold value, a working temperature range, a viscosity detection threshold value and a pressure gradient safety limit value, configuring differentiated process parameters according to liquid viscosity grades, configuring multistage pressure regulation stages and high-frequency vibration parameters by high-viscosity liquid, configuring medium-number pressure regulation stages and medium-frequency vibration parameters by medium-viscosity liquid, and configuring less pressure regulation stages and low-frequency vibration parameters by low-viscosity liquid.
3. 3. The adaptive vacuum degassing process optimization method based on the intelligent control system as set forth in claim 1, wherein S2 further includes: measuring rheological parameters of the liquid by a viscometer and a densimeter, and simultaneously analyzing the content and the size distribution of bubbles in the liquid by adopting a visual identification technology; acquiring a liquid surface image based on an industrial camera, identifying a bubble boundary by applying a deep learning target detection model, calculating a bubble size parameter by combining an image processing algorithm, and dividing the bubble into a plurality of size grades according to a bubble size range; Constructing a multidimensional characteristic vector model containing viscosity, density and bubble content, carrying out normalization processing on each dimensional parameter, and dynamically adjusting weight distribution of each parameter in the model according to the liquid type; Calculating Euclidean distance of current liquid characteristic parameter and record parameter in history database, i.e Screening multiple histories with smaller distance value, and applying weighted cosine similarity calculation to the screened histories And selecting a historical parameter combination with similarity higher than a preset threshold value as an initial defoaming strategy parameter.
4. 4. The adaptive vacuum defoaming process optimization method based on an intelligent control system according to claim 1, wherein loading a pre-trained network model in S3, generating a defoaming strategy by adopting a multi-objective optimization algorithm, and verifying the defoaming strategy in a virtual environment comprises: Extracting characteristics of space data of a liquid viscosity-time change curve and bubble size distribution, analyzing time sequence dependency relation of historical process parameters, loading a pre-training model, and carrying out parameter fine adjustment aiming at current liquid characteristics; Setting multi-target optimization indexes, wherein the multi-target optimization indexes comprise a defoaming efficiency index based on the reduction ratio of the number of bubbles, an energy consumption index based on the product of equipment power consumption and running time, and a product quality constraint condition for limiting the upper limit of bubble residual rate and the viscosity fluctuation range, the defoaming efficiency index is delta N/N_initial, delta N is the reduction amount of the number of bubbles, N_initial is the initial number of bubbles, the energy consumption index is sigma (pump_power+heater_power) x time, and pump_power is the power consumption of a vacuum Pump, and heater_power is the power consumption of a Heater; generating a plurality of groups of candidate strategies through an optimization algorithm, grading the strategies in a layering manner based on the pareto front theory, executing genetic operation on a high-level strategy set, and reserving the strategies with good optimization effect to enter iterative optimization; And inputting the generated deaeration strategy parameters into a virtual simulation environment for test verification, and triggering a strategy parameter adjustment mechanism when the microbubble secondary aggregation phenomenon is detected.
5. 5. An adaptive vacuum debubbling process optimization method based on an intelligent control system according to claim 1, wherein said executing a debubbling strategy comprising four successive process stages comprises: The first stage adopts a gradual depressurization control strategy, monitors the viscosity change of the liquid and dynamically adjusts the depressurization rate, simultaneously monitors the temperature change and adjusts the heating power, applies low-frequency vibration, and determines the duration of the stage according to the viscosity characteristic of the liquid; The second stage adopts a step-by-step depressurization control strategy, gradually reduces the system pressure according to preset time intervals, monitors the pumping speed of the vacuum pump, adjusts the working parameters of the pump, and applies mechanical vibration matched with the natural frequency of the bubbles; In the third stage, a multi-stage vacuum pump cooperative working mode is adopted to improve the vacuum degree, the temperature change of the liquid is monitored, a temperature rise compensation mechanism is triggered when abnormal temperature fluctuation is detected, and process temperature parameters are adjusted according to the viscosity characteristics of the liquid; And the fourth stage is to maintain the preset time under the final vacuum degree condition, then gradually restore to the normal pressure state according to the pressure recovery curve, continuously monitor the change of the number of bubbles, trigger an emergency depressurization program when the abnormal increase of the number of bubbles is detected, and execute on-line quality detection.
6. 6. The adaptive vacuum defoaming process optimization method based on the intelligent control system according to claim 1, wherein the step S5 of fusing multisource monitoring data based on a kalman filtering algorithm, analyzing the working condition state, and adjusting the defoaming process parameters comprises the following steps: The method comprises the steps of collecting pressure, temperature and vibration acceleration data in the defoaming process as state variables to construct a discrete time state transition matrix, calculating a sensor noise covariance matrix based on historical operation data to correct measured values, adjusting Kalman gain parameters according to sensor stability, reducing sensor weight and improving standby sensor weight when detecting that the sensor drift rate exceeds a preset standard, and outputting a state vector after multi-source data fusion; Carrying out short-term prediction on the pressure change trend based on Kalman filtering, and calculating the current liquid real-time viscosity value by combining a liquid viscosity-temperature relation curve stored in history; According to the deviation of the fusion pressure value and the target pressure value and the predicted pressure change rate, the working power of the vacuum pump is adjusted, and when the pressure is higher or the change rate exceeds a threshold value, the auxiliary vacuum pump is started; Calculating a predicted viscosity value according to the fusion temperature value, and correspondingly adjusting the power of the heater; And analyzing the distribution characteristics of the oscillation frequency of the bubble based on the fused vibration data, determining the optimal vibration frequency parameter, and triggering a vibration attenuation control program when the vibration amplitude exceeds a preset threshold.
7. 7. The adaptive vacuum defoaming process optimization method based on the intelligent control system according to claim 1, wherein the step S6 of detecting the bubble residual rate and counting the product consistency data, analyzing the relevance of the process parameters and the defoaming effect, and updating the decision strategy based on the reward signal generated by the defoaming effect comprises the following steps: acquiring a processed liquid surface image through an industrial camera, identifying and calculating the number of bubbles in a unit volume by applying a deep learning target detection model, acquiring bubble residual rate data, and carrying out statistical comparison on the current batch bubble residual rate and historical qualified batch data; Sampling and recording real-time data of pressure, temperature, vibration and viscosity in the defoaming process at a set frequency, adding a time stamp and a batch identifier to the data record, and constructing a metadata tag system of the production batch; Extracting key characteristic parameters affecting the defoaming effect from historical process data, constructing a characteristic matrix, and quantifying the influence weight of each process parameter on the defoaming result by using an interpretable machine learning method; Constructing a multidimensional rewarding function containing indexes of bubble residual rate, energy consumption and product consistency, generating a comprehensive rewarding signal, defining rewarding signal R=alpha (1-bubble residual rate) +beta (1-energy consumption/kWh) +gamma product consistency, wherein alpha and beta are corresponding weight parameters, optimizing a defoaming process parameter model based on the rewarding signal, and updating strategy network parameters through iterative training.

Description

Self-adaptive vacuum defoaming process optimization method based on intelligent control system Technical Field The invention relates to the technical field of defoaming process optimization, in particular to a self-adaptive vacuum defoaming process optimization method based on an intelligent control system. Background In the glass bonding process, vacuum defoamation is a key link for eliminating bubbles of an adhesive layer and guaranteeing bonding strength and optical consistency, and the prior art is dependent on a vacuum defoamation scheme with fixed parameters, and has limitations. On one hand, the traditional equipment lacks the perceptibility of characteristics such as liquid viscosity and temperature, so that the defoaming strategy cannot be matched with the differentiated requirements of high-viscosity liquid and low-viscosity liquid, microbubbles are often generated due to out-of-control pressure reduction rate or the energy consumption is too high, and on the other hand, the traditional system does not integrate a response mechanism of multi-source monitoring data, so that sudden pressure mutation or viscosity fluctuation in the defoaming process is difficult to cope with, and the risk of equipment damage is increased. In addition, the traditional quality evaluation only depends on a single bubble residual rate index, lacks the associated modeling capability of process parameters and defoaming effects, is difficult to realize strategy iterative optimization, severely restricts the yield and the production efficiency of the glass laminating process, and needs a self-adaptive vacuum defoaming solution integrating intelligent perception, optimization and closed-loop feedback. Disclosure of Invention The invention aims to provide an adaptive vacuum defoaming process optimization method based on an intelligent control system. The invention aims to solve the problems that the intelligent control system is used for realizing liquid characteristic self-sensing, strategy generation, multi-source data fusion regulation and control and closed loop feedback optimization, matching the defoaming requirements of liquids with different viscosities, synchronously reducing energy consumption and improving the bubble elimination rate and the process stability of a bonding layer. An adaptive vacuum defoaming process optimization method based on an intelligent control system adopts the following technical scheme: S1, before the deaeration operation is started, the system executes a comprehensive equipment inspection and parameter calibration program, and initial operation parameters are set according to the characteristics of the liquid to be processed; S2, comprehensively detecting liquid characteristics through a multidimensional sensing technology, matching with a preset characteristic database, and establishing a liquid characteristic vector model; s3, loading a pre-trained network model by a decision engine, combining the defoaming efficiency, the product quality and the energy consumption index, generating a preliminary defoaming strategy based on input characteristics by adopting a multi-objective optimization algorithm, and testing and verifying the generated defoaming strategy in a virtual environment; S4, the defoaming strategy comprises four process stages which are continuously executed, wherein the first stage is used for carrying out mild pre-defoaming treatment, the second stage is used for carrying out intensified defoaming treatment, the third stage is used for carrying out fine defoaming treatment, and the fourth stage is used for carrying out stabilizing treatment; S5, integrating multisource monitoring data based on a Kalman filtering algorithm, analyzing the current working condition state, adjusting parameters of a defoaming process according to feedback, simultaneously establishing a pressure-temperature safety guarantee mechanism, and presetting an automatic treatment scheme of various abnormal conditions; And S6, constructing a multi-index quality evaluation system to comprehensively evaluate the defoaming effect, detecting the bubble residual rate, counting the product consistency data, storing the complete defoaming process data, analyzing the relevance of the technological parameters and the defoaming effect, incorporating the successful cases into a knowledge base to optimize a decision model, updating a decision strategy according to a reward signal generated by the defoaming effect, and improving the model prediction accuracy through iterative adjustment. Further, the system in S1 executes a comprehensive equipment inspection and parameter calibration procedure, and sets initial operation parameters according to the characteristics of the liquid to be treated, including: Calibrating a vacuum system, detecting the working performance of a vacuum pump, and calibrating the measurement precision of a vacuum electronic pressure gauge and a vacuum electronic pressure regulating valve; Zero cal