CN-121979130-A - Thermodynamic process control method for glass needle tube for prefilled syringe

CN121979130ACN 121979130 ACN121979130 ACN 121979130ACN-121979130-A

Abstract

The invention discloses a thermodynamic process control method of a glass needle tube for a pre-filling injector, in particular to the technical field of medicinal glass production for the pre-filling injector, which comprises a multi-mode sensing configuration and calibration step of a pre-filling glass product, a multi-mode sensing fusion and thermodynamic state estimation step of the glass needle tube for the pre-filling injector, an on-line correction and working condition identification step of a digital twin model of the pre-filling glass product, a thermodynamic process self-adaptive control decision step of the pre-filling glass product, an execution and feedback step based on a double-stage high-temperature regulation and control process of the pre-filling glass product and a historical data archiving and self-learning optimization step of the thermodynamic process of the pre-filling glass product, the whole process of the pre-encapsulated glass product dual-stage high-temperature regulation and control process is finely and intelligently controlled through multi-source sensing data fusion, a digital twin model and a reinforcement learning control strategy linkage.

Inventors

ZHANG JUN
MA WENFENG
REN KEJU
WU SHUSEN
Lv Hansheng
LIU JIAN
YUAN HENGXIN

Assignees

山东省药用玻璃股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251229

Claims (8)

1. The thermodynamic process control method for the glass needle tube for the prefilled syringe is characterized by comprising the following steps of: the method comprises the steps of pre-potting glass product multi-mode sensing configuration and calibration, and executing a multi-mode sensor arrangement flow, a multi-mode sensor calibration flow and a multi-mode sensing time synchronization flow; A step of multi-mode sensing fusion and thermodynamic state estimation of a glass needle tube for a prefilled syringe, wherein a multi-mode data preprocessing flow, a multi-mode feature extraction flow, a multi-mode sensing fusion flow and a thermodynamic state estimation flow are executed; the method comprises the steps of carrying out on-line correction and working condition identification on a digital twin model of the pre-encapsulated glass product, and executing an initialization flow of the digital twin model, an on-line correction flow and a working condition identification flow of the digital twin model; a thermodynamic process self-adaptive control decision step of the pre-encapsulation glass product, wherein a control target constraint generation flow, a reinforcement learning control strategy reasoning flow and a control instruction issuing flow are executed; Based on the execution and feedback steps of the pre-encapsulated glass product double-stage high-temperature regulation process, executing a glass tube cutting execution flow, a primary high-temperature treatment execution flow, a directional rapid cooling execution flow, a secondary high-temperature melting execution flow, a mechanical breaking execution flow and an execution process on-line feedback flow; And (3) a pre-encapsulation glass product thermodynamic process historical data archiving and self-learning optimizing step, and executing a historical data archiving flow, a control strategy training flow and a control strategy version management flow.
2. The method for controlling the thermodynamic process of the glass needle tube for the pre-filled and sealed syringe according to claim 1, wherein in the steps of multi-mode sensing configuration and calibration of the pre-filled and sealed glass product, the multi-mode sensor arrangement process is characterized in that an infrared thermal image acquisition unit is arranged outside a corresponding area of a primary high-temperature processing execution process and a secondary high-temperature melting execution process, an optical fiber Bragg grating strain acquisition unit is arranged on a fixture structure close to a glass tube, an acoustic emission signal acquisition unit is arranged near the fixture and a mechanical fracture execution unit, a visual detection acquisition unit is arranged above a glass tube conveying and positioning unit, the multi-mode sensor calibration process is used for calibrating amplitude values of the acquisition units by using a standard glass tube and a standard working condition, and the multi-mode sensing time synchronization process is used for time alignment of the acquisition channels by using a unified clock.
3. The method for controlling the thermodynamic process of the glass needle tube for the prefilled syringe according to claim 1, wherein in the steps of multimodal sensing fusion and thermodynamic state estimation of the glass needle tube for the prefilled syringe, a multimodal data preprocessing flow performs gray-temperature conversion and spatial filtering on infrared thermal image data, bandpass filtering and temperature compensation on fiber bragg grating strain data, time-frequency conversion on acoustic emission signals, distortion correction on visual images, and multimodal feature extraction flow extracts temperature gradient features, strain distribution features, acoustic event features and geometric features respectively, a multimodal sensing fusion flow obtains fusion feature vectors by using a space-time coding network, and a thermodynamic state estimation flow outputs estimated temperature and stress field state quantities based on the fusion feature vectors and a digital twin model.
4. The method for controlling the thermodynamic process of the glass needle tube for the pre-filled and sealed syringe according to claim 1, wherein in the steps of on-line correction and working condition identification of the digital twin model of the pre-filled and sealed glass product, the digital twin model initialization process establishes a three-dimensional finite element model based on geometric parameters of the glass tube, thermophysical parameters of the material, thermal boundary conditions of a primary high-temperature treatment execution process, a directional rapid cooling execution process and a secondary high-temperature melting execution process, the digital twin model on-line correction process corrects parameters such as heat conductivity coefficient, convection heat conductivity coefficient and contact thermal resistance according to the deviation of the temperature and stress state quantity output by the thermodynamic state estimation process and a model prediction result, and the working condition identification process utilizes the current boundary conditions of the digital twin model and the characteristics output by the multi-mode sensing fusion process to determine the current process stage of the glass tube.
5. The method for controlling the thermodynamic process of the glass needle tube for the pre-encapsulation injector according to claim 1, wherein in the thermodynamic process self-adaptive control decision step of the pre-encapsulation glass product, a control target constraint generating process maps target temperature intervals, temperature gradient limits and stress limits of different stages into constraint ranges of flame power, water spraying flow and impact parameters of a mechanical fracture executing unit according to working condition labels given by a preset process specification and working condition identification process, a reinforcement learning control strategy reasoning process takes a state quantity output by the thermodynamic state estimating process and a predicted state output by a digital twin model on-line correction process as input, continuous control actions are generated under the constraint provided by the control target constraint generating process, and a control instruction issuing process converts the control actions into specific executing parameters and issues the specific executing parameters to a pre-encapsulation glass product thermal executing mechanism network element.
6. The thermodynamic process control method for the glass needle tube for the pre-filling and sealing injector is characterized in that in the executing and feedback steps of the double-stage high-temperature regulation and control process based on the pre-filling and sealing glass product, a glass tube cutting executing flow is cut after being positioned by a glass tube conveying and positioning unit, a primary high-temperature treatment executing flow heats a cutting area according to flame power set by a control instruction issuing flow when a working condition identification flow judges that the primary high-temperature treatment executing stage is carried out, a directional rapid cooling executing flow carries out directional spraying according to spraying flow set by the control instruction issuing flow when the temperature reaches a threshold value, a secondary high-temperature melting executing flow carries out secondary heating according to power and path set by the control instruction issuing flow when a thermodynamic state estimation flow judges that the temperature difference between a surface layer and a core layer meets the condition, and a mechanical breaking executing flow completes separation according to impact parameters set by the control instruction issuing flow when the working condition identification flow judges that the separation condition is met.
7. The method for controlling the thermodynamic process of the glass needle tube for the pre-filled and sealed syringe according to claim 1, wherein in the step of executing and feeding back the two-stage high-temperature regulation and control process based on the pre-filled and sealed glass product, the on-line feeding back process of the executing process continuously collects multi-mode data during the glass tube cutting executing process, the primary high-temperature treatment executing process, the directional rapid cooling executing process, the secondary high-temperature melting executing process and the mechanical breaking executing process, and the multi-mode data with the working condition label and the time stamp are written into the real-time data cache for the state update of the thermodynamic state estimating process and the control input of the reinforcement learning control strategy reasoning process.
8. The method of claim 1, wherein in the steps of pre-filling and sealing glass product thermodynamic process historical data archiving and self-learning optimization, the historical data archiving flow stores multi-mode data generated by an on-line feedback flow of an executing process and a control instruction generated by a control instruction issuing flow into historical data according to a batch number and a working condition label, the control strategy training flow is based on the data stored by the historical data archiving flow and invokes a digital twin model established by a digital twin model initialization flow and a model corrected by the digital twin model on-line correction flow to perform off-line training, and the control strategy version management flow carries out version management on strategy parameters output by the control strategy training flow and issues a selected version to a reinforcement learning control strategy reasoning flow in a shutdown window.

Description

Thermodynamic process control method for glass needle tube for prefilled syringe Technical Field The invention relates to the technical field of medicinal glass production for pre-filled and sealed syringes, in particular to a thermodynamic process control method for a glass needle tube for a pre-filled and sealed syringe. Background In the traditional manufacturing process of pre-encapsulated glass products, a linear process chain of cutting, heating, cooling and mechanical breaking is generally adopted, namely glass tube cutting is carried out on a glass tube through a glass cutter, then a natural gas flame is utilized to heat a cutting area at high temperature, then a water spray nozzle is utilized to rapidly cool a local area, and finally a small hammer is utilized to strike the glass tube to break the glass tube, so that segmentation is realized. Aiming at the problem that the defects of glass scraps stay high for a long time, the prior art provides a double-stage high-temperature regulation and control process on the basis, the thermodynamic path between original high-temperature heating and mechanical fracture is reconstructed into a sequence of 'glass tube cutting-primary high-temperature treatment-directional rapid cooling-secondary high-temperature melting-mechanical fracture', partial cutting stress is released through primary high-temperature treatment, the stress field and microcracks near the section are regulated by combining directional rapid cooling and secondary high-temperature melting, and meanwhile, the cooperative control of heating, cooling and melting time sequences is realized by matching with certain temperature monitoring and dynamic temperature control, so that the glass scraps generation risk is relieved on the process level. The scheme based on the dual-stage high-temperature regulation and control process still has various defects in a process control layer, namely, firstly, a plurality of schemes only arrange single or small quantity of sensors such as temperature, strain or acoustic emission at local key stations, lack of systematic multi-mode sensing configuration and calibration steps of pre-encapsulation glass products, have non-unification of time reference and space coordinates among multi-source data, and are difficult to support multi-mode sensing fusion in a true sense, secondly, although the prior art proposes to analyze thermal stress distribution by utilizing a simulation model, a digital model mainly adopts offline analysis, on-line correction and working condition identification steps of a pre-encapsulation glass product digital twin model which is continuously coupled with field data are not formed, and lack of fine on-line identification and prediction of thermodynamic states of glass tubes in a glass tube cutting execution flow, a primary high-temperature processing execution flow, a secondary high-temperature melting execution flow and a mechanical breaking execution flow, thirdly, the prior art control is multi-reliant on preset process curves, experience parameters and conventional feedback control, lack of system utilization of a large quantity of historical data accumulated in the on-line feedback flow, the pre-encapsulation glass products and the pre-encapsulation glass products which are not established, the self-archive process data and the self-learning strategy is difficult to realize, and the self-adaption to be realized, and the self-adaption to the self-adaption process window is difficult to realize, and the self-adaption is still required to be realized, and stable, and the self-adaption is required to test and stable. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides a thermodynamic process control method for a glass needle tube for a prefilled syringe, which solves the problems in the prior art through the following scheme. In order to achieve the purpose, the invention provides the following technical scheme that the thermodynamic process control method for the glass needle tube for the prefilled syringe comprises the following steps: the method comprises the steps of pre-potting glass product multi-mode sensing configuration and calibration, and executing a multi-mode sensor arrangement flow, a multi-mode sensor calibration flow and a multi-mode sensing time synchronization flow; A step of multi-mode sensing fusion and thermodynamic state estimation of a glass needle tube for a prefilled syringe, wherein a multi-mode data preprocessing flow, a multi-mode feature extraction flow, a multi-mode sensing fusion flow and a thermodynamic state estimation flow are executed; the method comprises the steps of carrying out on-line correction and working condition identification on a digital twin model of the pre-encapsulated glass product, and executing an initialization flow of the digital twin model, an on-line correction flow and a working condition identification flow of the digital twin model; a the