CN-122022605-A - Automatic production quality control system and method for diaphragm boxes of diaphragm changing instrument

CN122022605ACN 122022605 ACN122022605 ACN 122022605ACN-122022605-A

Abstract

The invention belongs to the field of automatic production control and quality detection, and particularly relates to an automatic production quality control system and method for a diaphragm changing instrument diaphragm box. The system comprises a multidimensional collaborative data acquisition module, a feature depth extraction and fusion module, an intelligent defect identification and classification module, a process parameter dynamic compensation module and a full-quantization quality tracing unit. The multi-mode quality characteristics of the bellows are collected and fused in real time, the defect grade is judged by using a deep learning model, the process deviation is calculated based on the identification result, and a correction instruction is automatically issued to an executing mechanism, so that closed-loop adjustment of production parameters and full life cycle tracing of quality are realized. The method improves the identification precision of the defects of the diaphragm capsule and the degree of production automation, and effectively reduces the production cost.

Inventors

Cao pan
ZHANG QUAN
SUN XIN
ZHI KANGBO
WANG JINGLEI
XU HAIBO
FANG YUYU
FANG LIHONG

Assignees

杭州微智兆智能科技有限公司
四川省生态环境监测总站

Dates

Publication Date: 20260512
Application Date: 20260415

Claims (10)

1. An automatic quality control system for the production of film boxes of film changing instruments, comprising: The multi-dimensional collaborative data acquisition module is used for setting a plurality of monitoring nodes on a production line and acquiring multi-mode original data of the capsule in different procedures in real time; The multi-mode original data comprise high-resolution optical images on the surface of the diaphragm, fluorescence excitation spectra of the packaging glue path, ultrasonic reflection signals of the welding position of the shell and differential pressure change data of the packaging cavity; The feature depth extraction and fusion module is used for receiving the original data output by the multidimensional collaborative data acquisition module, carrying out quantitative feature recognition on membrane wrinkles, glue path cracks, shell cracks and impurity particles by utilizing a preset feature extraction algorithm, and constructing a multidimensional quality feature vector; The intelligent defect recognition and classification module is used for inputting the quality feature vector into a pre-constructed deep learning recognition model, judging the quality grade of the current capsule by carrying out nonlinear mapping and weight matching on the feature vector, and outputting a recognition result comprising the defect type, the defect position and the defect severity; the process parameter dynamic compensation module is used for calculating the deviation association degree between the current production process parameter and the target quality index based on the identification result of the intelligent defect identification and classification module, generating a correction instruction for an execution mechanism of the production equipment according to the deviation association degree, and realizing real-time closed-loop adjustment of dispensing pressure, welding energy and packaging travel; And the full-quantization quality tracing unit is used for storing the unique identification code of each capsule, the corresponding multi-mode original data, the identification result and the process compensation record in a correlated way, and constructing a production quality file covering the full life cycle.
2. The automated quality control system for a membrane changer capsule of claim 1, wherein the multi-dimensional collaborative data acquisition module comprises: The precise visual monitoring assembly is arranged at a diaphragm laying station and comprises an industrial camera with more than 1200 ten thousand pixels, a telecentric lens and a multi-angle light-emitting diode annular light source, and is used for capturing micron-level fluctuation information of the surface of the diaphragm; The fluorescence spectrum analysis assembly is arranged behind the dispensing packaging station and comprises an ultraviolet excitation light source with 365 nm wavelength, a high-pass filter and a spectrum sensor, and is used for acquiring fluorescence distribution characteristics of a glue path; the ultrasonic nondestructive testing assembly is integrated in the welding press head and is used for transmitting ultrasonic waves of 20-40 kilohertz to the welding interface and collecting waveform distortion rate of the reflected waves; The high-precision differential pressure sensing assembly is arranged at a finished product detection station and comprises a differential pressure transmitter with the measuring range of 0 Pa to 1000 Pa and the precision of 0.1 Pa, and is used for monitoring the pressure attenuation value of the inside of the bellows within 1 second to 3 seconds after the bellows is filled with preset pressure air.
3. The automated manufacturing quality control system for film changing instrument bellows of claim 2, wherein the precision vision monitoring assembly further comprises a light source controller and an exposure trigger; the light source controller and the exposure trigger realize millisecond synchronization so as to ensure that the spatial resolution of the captured optical image of the surface of the diaphragm is better than 20 micrometers when the conveyor belt is in a high-speed moving state of 0.5 to 1.5 meters per second; the irradiation angle of the ultraviolet excitation light source is set to be 45 degrees with the surface of the capsule so as to inhibit the specular reflection interference of the surface of the shell.
4. The automatic production quality control system for membrane boxes of a membrane changer according to claim 2, wherein the high-precision differential pressure sensing assembly adopts a symmetrically distributed double-cavity structure; one side of the double-cavity structure is a reference standard cavity, and the other side is a bellows cavity to be detected; The high-precision differential pressure sensing assembly counteracts measurement errors generated by environmental temperature fluctuation by comparing pressure difference values between the reference standard cavity and the capsule cavity to be measured.
5. The automated quality control system for membrane changer bellows of claim 1, wherein the feature depth extraction and fusion module performs the following logic when extracting feature parameters: Extracting texture distribution and roughness quantization indexes of the surface of the diaphragm by carrying out characteristic descriptor operation on the high-resolution optical image of the surface of the diaphragm; The method comprises the steps of (1) extracting parameters representing glue path uniformity and solidification degree by carrying out space distribution variance calculation and spectrum peak value offset monitoring on a fluorescence excitation spectrum of a packaging glue path; the method comprises the steps of performing fast Fourier transform on ultrasonic reflection signals of a welding position of a shell, and extracting energy attenuation proportion of a high frequency band in a frequency domain; and mapping the characteristic parameters of different physical dimensions to a unified numerical interval by adopting a linear weighting and normal normalization combination algorithm to generate a high-dimensional characteristic vector.
6. The automated quality control system for film changer bellows of claim 1, wherein the intelligent defect identification and classification module, when performing a classification decision, follows the following criteria: when the highest probability value output by the activation function exceeds 95%, directly issuing a corresponding quality grading result; and when the highest probability value is between 70% and 95%, judging that the current capsule is a suspected defect product, and triggering an artificial secondary verification instruction to a quality inspection terminal.
7. The automated quality control system for a film changer bellows of claim 1, wherein the process parameter dynamic compensation module comprises: the association degree analysis unit is used for distinguishing random interference and systematic process drift by monitoring defect distribution of more than 10 unqualified products in the production sequence; the correction compensation algorithm unit is used for driving the dispensing pump to increase the driving pressure according to the stepping amount of 1-5% if the width of the glue path is continuously narrowed when the systematic process drift is identified, and increasing the ultrasonic welding energy according to the amplitude of 50-200 joules if the welding position strength is insufficient; And the feedback verification unit is used for extracting the detection data of the regulated and controlled working procedure to carry out closed-loop verification in the next production cycle after the correction instruction is executed.
8. The automated quality control system for a film changer bellows of claim 7, wherein the process parameter dynamic compensation module performs a collision detection procedure prior to issuing a correction command; The conflict detection program is used for comparing the technological parameters to be adjusted with the physical limit of the production equipment and a preset safe operation interval, and refusing to execute compensation and sending a manual intervention request when the parameters to be adjusted exceed the safe interval.
9. The automatic production quality control system for membrane changing instrument membrane boxes according to claim 1, wherein the full-quantification quality traceability unit is connected with a programmable logic controller of a production site and a manufacturing execution system of an enterprise through an industrial Ethernet; The system operates in a three-layer network architecture and comprises a field control layer for executing production actions and rejection actions, an edge calculation layer for operating a feature extraction algorithm and a deep learning identification model, and a cloud management layer for storing long-term quality files and performing cross-batch statistical analysis.
10. A production automation quality control method for membrane changing instrument capsules, realized based on the production automation quality control system of any one of claims 1 to 9, characterized by comprising the following steps: Step 1, initializing a system, namely loading optical, spectral, acoustic and pressure reference parameters preset for a standard capsule sample, and executing self calibration on a sensing assembly; step 2, synchronously acquiring data, namely synchronously acquiring three-dimensional shape data of a diaphragm, a glue path fluorescence spectrum, a welding interface echo waveform and cavity differential pressure attenuation data in the process of diaphragm box circulation; Step 3, feature fusion processing, namely denoising and enhancing the acquired original signals, extracting physical features and fusing comprehensive fingerprint feature vectors representing the quality of the monomer bellows; step 4, intelligent defect identification, namely inputting the comprehensive fingerprint feature vector into an identification model, judging that the membrane is wrinkled if the flatness deviation of the membrane exceeds 0.05 mm, judging that the glue path is broken if the continuous missing length of the fluorescence intensity of the glue path exceeds 0.5mm, judging that the welding is cold-welded if the energy leakage proportion of ultrasonic reflection signals exceeds 15%, and judging that the air tightness is unqualified if the differential pressure attenuation value of the cavity exceeds 5 Pa within 1 second; Step 5, closed loop feedback compensation, namely issuing a process adjustment instruction to dispensing and welding equipment according to a defect identification result, and driving a sorting mechanism to reject waste; and step 6, archiving and optimizing, namely uploading the detection data and the compensation record to a distributed storage architecture, and performing reinforcement training on the identification model by periodically utilizing the accumulated data.

Description

Automatic production quality control system and method for diaphragm boxes of diaphragm changing instrument Technical Field The invention belongs to the field of automatic production control and quality detection, and particularly relates to an automatic production quality control system and method for a diaphragm changing instrument diaphragm box. Background In the fields of precision instrument manufacturing and laboratory automation, the production quality of core consumables of various high-precision tip experimental equipment directly determines the accuracy of experimental analysis and the operation stability of a system. Along with the rapid development of the industries of biological medicine, environmental monitoring and material science, higher requirements are put forward on the production consistency and quality monitoring of a precise filter assembly and a carrier film box thereof, and the strict quality control management of the whole production process by an automatic technical means becomes a core link of the upgrade of the precise manufacturing industry. A diaphragm capsule for a diaphragm changing instrument relates to multi-working procedures such as micro diaphragm positioning, high polymer shell packaging, precise sealing and the like, and an automatic production quality control system aims at realizing real-time monitoring and verification of the appearance structure, the diaphragm flatness and the sealing performance of the diaphragm capsule by integrating a multi-source acquisition unit and an on-line monitoring algorithm. The operation efficiency of such systems is directly related to the mechanical co-efficiency of the membrane changer in performing an automated membrane change operation and the fluid safety in subsequent filtration operations. The prior art mainly faces the problems of unbalanced degree of automation, limited detection precision and the like in the quality control process of the diaphragm capsule. The traditional detection mode is often focused on off-line spot inspection or single airtight performance test, lacks synchronous identification capability of composite defects such as membrane wrinkles, gum path overflows, shell microcracks and the like, is limited by high-speed production beats, and is difficult to realize multi-dimensional full quality control on the premise of ensuring productivity. In addition, the data correlation between the existing quality control system and the production control level is insufficient, so that the detected quality defects cannot be fed back to the process parameter end in real time to be dynamically compensated, and abnormal fluctuation in the production process is difficult to intercept in time. The discretization and hysteresis of the quality control means not only increases the risk of leakage or film clamping of the film box in the use process, but also obviously limits the reliability and compliance of the integral operation of the film changer. Disclosure of Invention The invention provides an automatic quality control system and method for the production of a diaphragm changer, which are used for solving the problems of single detection dimension, unbalanced automatic quality control degree, limited defect identification precision and missing of a closed loop of quality control data and process parameter feedback in the diaphragm production process in the prior art, and further realizing high-precision, full-quantization and real-time dynamic monitoring on the appearance structure, the diaphragm flatness and the sealing performance of the diaphragm. The technical scheme of the invention is to provide a production automatic quality control system for a membrane replacement instrument membrane box, which comprises: The multi-dimensional collaborative data acquisition module is used for setting a plurality of monitoring nodes on a production line and acquiring multi-mode original data of the diaphragm capsule in different working procedures in real time, wherein the multi-mode original data comprises a high-resolution optical image of the surface of the diaphragm, a fluorescence excitation spectrum of a packaging glue path, an ultrasonic reflection signal of a shell welding position and differential pressure change data of a packaging cavity. The feature depth extraction and fusion module is used for receiving the original data output by the multidimensional collaborative data acquisition module, carrying out quantitative feature recognition on membrane wrinkles, glue path cracks, shell cracks and impurity particles by utilizing a preset feature extraction algorithm, and constructing a multidimensional quality feature vector. The intelligent defect recognition and classification module is used for inputting the quality feature vector into a pre-constructed deep learning recognition model, judging the quality grade of the current capsule by carrying out nonlinear mapping and weight matching on the feature vector, and output