No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Trustworthiness of machine learning models in manufacturing applications using the example of electronics manufacturing processes
Abstract
The design phase of products decides about a large portion of their costs. Both applied manufacturing technologies and systems are ultimately defined by the products that have to be assembled. Due to global trends towards e-mobility and regenerative energy, electronic printed circuit boards with high copper content and mixed SMT/THT assembly is the predominant assembly scenario. This leads to thermally challenging soldering operations in THT-soldering as high copper masses dissipate the required heat and no reliable manufacturability check is available. Consequently, in electronic manufacturing complex thermal processes have still to be defined in iterative experimental steps. Reliable models have great potential to reduce time-to-market and scrap during new product introduction. Future manufacturing systems require infrastructure that allows manufacturability checks in the early design phase and automatically provides process programs on the shop floor once the design is finished. However, the lack of availability, long-term experience and consequently trust in automatically, by artificial intelligence generated and optimized process programs, blocks the realization and hinders further productivity gain. In this paper, an ML audit methodology is suggested that mitigates this lack of trust by providing physically plausible prediction series and quantitative risk assessment to the user. This complements typical evaluation scores and adds the model behavior in the process context and transmits a sense of transparency. A neural network is trained with an augmented dataset, exemplary for the prediction of the hole fill in mini wave soldering process. The model is then used to predict series of predictions for increasing soldering time and temperatures comparable to design of experiments. The resulting time versus hole fill plots helps the operator understanding the behavior of the model and thus, demonstrate the potential benefit of this approach to the entire manufacturing site.
... In the process of soldering, two metal surfaces are joined together by first being heated to a temperature at which they melt and flow together, and then being cooled to produce a connection that is stable and long-lasting. It is utilized in a variety of techniques that are utilized in the production of electronic goods (Seidel et al., 2022). According to Khader and Yoon (2021), one of the most prevalent methods utilized in surface mount technology (SMT) is reflow soldering. ...
The world of manufacturing is undergoing a significant transformation, as the integration of artificial intelligence (AI) and machine learning techniques become more prevalent. In the realm of soldering, these advanced technologies are revolutionizing processes, enhancing efficiency, and optimizing precision in various applications. "Smart Soldering: Exploring the Synergy of Artificial Intelligence and Soldering Technologies" delves into the intricate relationship between AI and soldering in the manufacturing industry, showcasing groundbreaking studies that employ AI methodologies for different soldering-related challenges.
From defect detection in solid oxide fuel cells to predicting shear strength in microbump interconnections, AI techniques are proving invaluable in addressing complex issues and improving quality control. In the realm of electronics manufacturing, AI-driven methodologies have been applied to optimize processes, such as the mini wave soldering process and pin-in-paste technology, demonstrating potential benefits across entire manufacturing sites.
Furthermore, the integration of AI into materials science has opened up new avenues for understanding the complex interactions between materials and predicting their performance. Reinforcement learning, a subset of AI, has also emerged as a powerful tool for real-time optimization of manufacturing processes, as seen in stencil printing applications.
"Smart Soldering" brings together a collection of studies that epitomize the potential of AI and machine learning in soldering applications, offering readers a comprehensive understanding of the transformative impact these technologies are having on the manufacturing industry. By examining the latest research and breakthroughs, this book highlights the immense potential of AI in advancing soldering technologies and processes, ultimately paving the way for a smarter and more efficient manufacturing future.
The aim of this research is to develop a novel machine learning approach for fault detection in power electronics circuit boards used in textile mills. Power electronics circuit boards play a critical role in the smooth operation of textile mills. However, the failure of these boards can cause significant downtime and loss of productivity, resulting in substantial financial losses for the textile industry. Therefore, developing an efficient and accurate fault detection system for power electronics circuit boards is of utmost importance. The proposed approach employs machine learning algorithms to detect faults in real-time and mitigate the risk of downtime. The approach utilizes several techniques, such as signal processing, feature extraction, and classification, to analyze the power electronics circuit board's behavior and detect faults before they cause any significant damage. The machine learning models used in the proposed approach are trained using a vast dataset of power electronics circuit board signals and fault data, which are collected from various textile mills. Additionally, the research will evaluate the proposed approach's ability to detect faults in real-time, which is crucial for minimizing downtime and maximizing productivity in textile mills. The outcomes of this research will benefit the textile industry by reducing downtime, minimizing production losses, and improving the overall efficiency and productivity of textile mills.
Higher functional integration and increased electrical and reliability requirements are leading to increasingly complex printed circuit board (PCB) designs. Highly integrated PCBs with mixed SMT/THT assembly cause difficulties in soldering processes due to the thermal mismatch between heat demand and heat supply during soldering and the limited accessibility of THT-solder joints. This directly affects the manufacturability and robustness of the THT-soldering process with respect to sufficient hole fill according to typical quality standards like IPC-A-610. Partial hole fill also affects the mechanical reliability and fatigue resistance of a solder joint as well as its current carrying capacity and high frequency behavior. In the literature, several publications show the positive correlation between a larger hole diameter and vertical hole filling. In this paper, the fundamental relationships between hole diameter, hole filling and process robustness are investigated using design of experiments (DoE). It is shown that hole filling increases with increasing hole diameter, while keeping all other parameters constant. It was found that the process also becomes more robust at shorter solder contact times and lowers preheat temperatures. From these results, a general recommendation for larger THT holes in selective wave soldering can be derived. This is consistent with several other literature reports. In contrast to previous publications, it is not only shown how increasing the gap favors hole filling, but also what this means for the robustness of the soldering process to different PCB copper layer designs and soldering parameters. In addition, the impact of hole diameter is evaluated with respect to robotic component placement and back-end processes such as testing. A Gap-Ratio of 30–40 % is recommended for selective wave soldering to ensure robust soldering processes.
Unterlagen zur Vorlesung "Thermisches Management in der Leistungselektronik" an der Universität Erlangen-Nürnberg
A black-box model of a system is one that does not use any
particular prior knowledge of the character or physics of the
relationships involved. It is therefore more a question of
“curve-fitting” than “modeling”. In this
presentation several examples of such black-box model structures will be
given. Both linear and non-linear structures are treated. Relationships
between linear models, fuzzy models, neural networks and classical
non-parametric models are discussed. Some reasons for the usefulness of
these model types are also given. Ways to fit black box structures to
measured input-output data are described, as well as the more
fundamental (statistical) properties of the resulting models
Although the surface mount technology has accounted for 90% in the electronic products, the traditional wave soldering technology is still used in the future. This paper mainly discusses the wave soldering technology, focus on the difficult soldering technology of single row of DIP pins with pitch 1.27 mm, and QFP with pitch 0.5 mm. The special soldering pads were designed for the two kinds of components. The normal wave soldering process parameters combining with special coating process were used to solder the experimental board, and the result is what we expected. We get the optimal pad design for QFP with pitch 0.5 mm and the good pad design for single row of DIP pins with pitch 1.27 mm based on the experiment. At the same time, the experimental result shows that the special soldering pad design is needed when QFP and DIP are soldered by wave soldering.
PTH often suffers poor hole fillet in wave soldering on OSP PCB. When PCB thickness reaches 2.7mm in server products and when the process is of lead-free, this issue has become more and more prominent. This paper studies how to achieve a good hole fillet by a good design of PTH hole. Two main factors are discussed here - GND layer and Gap between the hole and the component pin. With minimum GND layers under electrical requirement and gap ratio of 15~20% (gap between PTH wall and PTH component pin is of (15-20%) *2.70mm, i.e. 0.40~0.54mm), a good hole fillet can be guaranteed. Further study also shows that such process parameters as flux volume and dipping time can be optimized to enhance wave solder capability and PCBA reliability when GND layer is reduced and hole diameter is enlarged. Further study also shows that flux type ought to be of high activity in dealing with 2.7-mm-thick PCB with surface finish of Enthone HT OSP type.
Pb-Free Selective Wave Solder Guidelines for Thermally Challenging PCBs
- R Mendez
- H Lowe
- I Marin
- C Monterrey