Lead bonding quality prediction control method based on machine learning

Abstract

The invention belongs to the technical field of electronic packaging lead bonding processes, and discloses a machine learning-based lead bonding quality prediction control method, which comprises the following steps of: selecting a plurality of key process parameters of lead bonding as key influence factors of lead bonding, and constructing a quality prediction neural network model based on the key influence factors; collecting real-time process parameters corresponding to the key influence factors when the lead bonding machine works; according to the real-time process parameters, the quality prediction neural network model is used for conducting lead bonding quality prediction, and a quality prediction result is obtained; and according to the quality prediction result, carrying out real-time adjustment on the process parameters of the lead bonding machine so as to ensure the expected bonding quality. According to the method, the product yield can be effectively increased, the influence of different machining working conditions on the product quality is effectively overcome, and the refined requirement for the product quality is met.

Description

technical field [0001] The invention belongs to the technical field of wire bonding technology for electronic packaging, and more specifically relates to a machine learning-based predictive control method for wire bonding quality. Background technique [0002] Since its invention, wire bonding has become the most common and widely used connection method for connecting wires between chips in the electronic packaging process due to its relatively low cost, easy process implementation, and wide application range. In order to ensure the smooth implementation of chip and external input and output, wire bonding usually needs to ensure smooth electrical connection between the chip and the outside, and the quality of wire bonding is directly related to the operational stability and product quality of electronic circuits. Since wire bonding is usually performed at a relatively small level, when external factors fluctuate, it is difficult to accurately perform manual identification an...

AI Technical Summary

Problems solved by technology

Since wire bonding is usually performed at a relatively small level, when external factors fluctuate, it is difficult to accurately perform manual identification and timely adjustment, which may easily lead to problems such as open circuit, short circuit, and unstable resistance of the bond due to the decline in bonding quality. These problems will directly affect the use effect and service life of the product

[0003] The quality of wire bonding is affected by many factors, such as bonding time, bonding force, bonding temperature, etc.; the existing quality detection methods are mainly offline monitoring methods, including pull test, push ball test, etc., and the detection feedback is slow and slow. It is destructive, so it is impossible to test all the leads. Most of the leads can only be inspected manually. The detection is cumbersome and the quality cannot be effectively guaranteed. At the same time, this detection process cannot find out the problems in the bonding. Manual experience judgment has high requirements for experience and inevitably leads to quality and efficiency problems caused by wrong decision-making

Images