Deep learning-based glue residue prediction method for time-pressure dispensing system

Abstract

The invention relates to a deep learning-based glue residue prediction method for a time-pressure dispensing system, and the method comprises the steps: constructing a deep learning model for predicting the glue residue of a glue storage tube of the time-pressure dispensing system, the input of the deep learning model comprises response air pressure sequence data and current air source air pressure data at an outlet of an electromagnetic valve during dispensing, and the output of the deep learning model is a colloid allowance predicted value; collecting a data set required for training the deep learning model, and training the deep learning model; and using the trained deep learning model to carry out colloid allowance. Compared with the prior art, the method has the advantages of simple steps, high prediction accuracy, easiness in application and popularization in engineering and the like.

Description

technical field [0001] The invention relates to a method for predicting the colloid residual of a time-pressure dispensing system, in particular to a method for predicting the residual colloid of a time-pressure dispensing system based on deep learning. Background technique [0002] Time-pressure dispensing is widely used in integrated circuit packaging and surface mount because of its simple structure and easy equipment maintenance. It extrudes the glue in the glue storage tube through compressed gas, coats the surface of the workpiece, and controls the amount of glue by adjusting the opening and closing time of the solenoid valve. But its disadvantage is that it is difficult to ensure the high precision and high stability of the glue output. For a long time, there has been research on the modeling of the time-pressure dispensing system. The mechanism modeling of the dispensing process is carried out, and the automatic control algorithm is used to ensure the stability of t...

AI Technical Summary

Problems solved by technology

However, the gas response process in the trachea is complex, and it is difficult to model the mechanism. There is a literature (Shen Zhengxiang, Dispensing control system and its performance control research in microelectronic packaging, 2005, Huazhong University of Science and Technology. p. 80) through regression or approximation methods, Construct the empirical function of the outlet pressure of the solenoid valve with respect to time under different colloid levels of the storage tube, so as to predict the margin in the project. However, due to the high degree of nonlinearity of the dispensing system, it is difficult to fit it well, and a set of empirical functions only It can realize the margin prediction under fixed dispensing air pressure, but it cannot meet the needs of working under different dispensing air pressures in engineering; there are also studies (Zhao Yixiang, modeling, control and realization of dispensing system for semiconductor packaging, 2004, Huazhong University of Science and Technology .page 101) Estimate the colloid residual value from the derivative of the current test tube cavity pressure Pc, that is, the slope. Although the residual volume prediction under multiple dispensing air pressures can be realized, the accuracy is low, and the test tube cavity pressure Pc is also low. Not easy to obtain in engineering

Images