Search

This study aims to identify and maintain optimal stencil printer settings in dynamic surface mount technology (SMT) assembly lines to control the solder paste volume transfer efficiency (TE) and increase the yield rates of surface mount assembly (SMA) lines. Stencil printing process (SPP) is a crucial procedure that determines the first pass yields in SMA lines. Stencil printing speed (PS) and printing pressure (PP) are two critical SPP parameters in the manufacturing process; identifying the optimal printer settings in complex fabrication environment is cost efficient but challenging. To search for the global optimal PS and PP values in real time with minimal computational efforts, a dynamic optimization SPP optimization system is proposed based on a novel guided evolutionary search (GES) strategy. The GES algorithm is flexible to adjust the feasible solution region based on real-time optimization results to maximize the probability of finding optimal solutions and minimize the computational burden. As a major extension on the classical evolutionary search, the GES minimizes the possibility to fall in the local optimal due to insufficiencies in the prediction accuracy and guarantees global optimality within the dynamically varying printing environment.

This article aims to propose a predictive abnormality detection model in the stencil printing process (SPP). The SPP is the main contributor to surface mounting technology (SMT) soldering defects. The prediction of abnormal conditions is necessary to enhance the first-pass yield and reduce the reworking costs of the printed circuit board (PCB) assembly line. In this research, a novel multiphase intelligent abnormality prognosis (IAP) framework is proposed. The model comprises two phases: the abnormality detection phase and the abnormality prediction phase. The first phase is to develop the random forest-based exponential weighted moving average (RF-based EWMA) control chart. The goal is to properly monitor the highly autocorrelated SPP process and effectively recognize the existing patterns. In the second phase, the accurate prediction of anomalies within the SPP before they arise is achieved. The integration of adaptive boosting (AdaBoost) predictive modeling and a moving recognition window approach is proposed. To discriminate the different patterns from each other, features are extracted using the sliding window, and then, the AdaBoost model is adopted to predict the occurrence of abnormal patterns in the SPP. The experimental results confirm the effectiveness and reliability of the proposed framework in early and accurate prediction of abnormal patterns within the SPP process to prevent solder paste printing defects and reduce the high reworking costs for large-scale production.

This article aims to develop a dynamic optimization model performing real-time control of a stencil printing process (SPP) by maintaining the optimal printer parameter settings. In a surface mount technology (SMT) assembly line, stencil printing is a major process that affects the yield of printed circuit boards (PCBs). During printing, environmental changes may induce the PCB’s printing results to deviate from initial optimal outcomes. To consistently improve the system performance, a real-time adaptation of the printer settings is an effective and cost-efficient approach. This research proposes a hybrid online optimization model by using online learning to predict real-time SPP volumes and an evolutionary search (ES) technique to determine the optimal settings. The prediction model investigates the printing volumes’ transfer efficiency (TE) in averages and standard deviations (SDs) with relevant features. From the model selection of the online-based learning, the multi-layer online sequential extreme learning machine (MOSELM) shows outstanding prediction performance with $R^{2}$ values of 97% for volume averages and 81% for SDs. From the real implicational results, the system achieves a $C_{pk}=2.8$ , outperforming other advanced models. The proposed framework exhibits a good balance between accuracy and retraining efficiency, promising effective SMT assembly dynamic control.

This research proposes a novel genetic algorithm-based online gradient boosting (GAOGB) model for incremental breast cancer (BC) prognosis. The development of clinical information collection technologies has brought in increasingly large amounts of stream data for BC research. Traditional batch learning models have shown limitations in: (1) real-time prognosis accuracy from losing the information of incremental changes of a patient’s pathological condition by time; (2) high redundancy due to the time required to retrain models every time new data are received. Online boosting is an efficient technique for learning from data streams. However, difficulties in parameter assignment and the lack of adaptiveness for batch learning base learners can degrade the performances of typical online boosting algorithms. The main objective of this research is to propose an incremental learning model for BC survivability prediction. To render a boosting algorithm with superiority in global optimal parameters, the genetic algorithm (GA) is integrated to an online gradient boosting scenario at the parameter selection phase, enabling real-time optimization. To enhance adaptiveness, an adaptive linear regressor is adopted as the base learner with minimal computational efforts, and updated in symphony with the online boosting model. The proposed GAOGB model is comprehensively evaluated on the U.S. National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program breast cancer dataset in terms of accuracy, area under the curve (AUC), sensitivity, specificity, retraining time, and variation at each iteration. Experimental results show that the proposed GAOGB model achieves statistically outstanding online learning effectiveness. With a highest 28% improvement on testing accuracy over its base learners, outperforming current state-of-art online learning methods, and approximating batch learning boosting algorithms, the GAOGB algorithm validates the impact of parameter, adaptiveness and convergence in devising practical online learning algorithms. The proposed GAOGB model demonstrates potential for practical incremental breast cancer prognosis, promising a combination of training effectiveness and efficiency.

This research proposes a stencil cleaning decision-making model in surface mount technology. Stencil cleaning is a critical process that influences the quality and efficiency of printing circuit boards. Stencil cleaning operation depends on various process variables, such as printing speed, printing pressure, and aperture shape. The objective of this research is to develop an intelligent model to guide stencil cleaning decision-making to reduce process defects. The stencil cleaning process is considered as a sequential detection problem in this study. Based on quality measures of printed historical boards, such as solder paste volume and the number of defects, a novel feature space is proposed by considering both short-term and long-term process trend. A gradient boosting model is applied to make the stencil cleaning decision. To validate the effectiveness of the proposed model, different scenarios are designed in the experimental test. State-of-art data mining models are also compared to the proposed cleaning decision-making model. Experimental results show that the proposed boosting-based intelligent model outperforms other models and can effectively provide the cleaning suggestion even the board design is changed in the future.

This research aims to develop a dynamic prediction model to assist real-time decision making of a stencil printing process by maintaining high prediction accuracy of the printing process. In a Surface Mount Technology (SMT) assembly line, the stencil printing process (SPP) accounts for more than 50% of the defectiveness of printed circuit boards (PCBs). During the printing process, environmental changes such as humidity or wear of blades may induce the PCBs printing results to deviate from the target volume. Thus, real-time adjustment of the printer settings (e.g., printing parameters, clean cycles, etc.) based on prediction of printing volumes is critical to maintain a high printing performance. However, research has been limited in real time SPP control, which is partially due to the difficulties in predicting the paste volumes with high accuracy and time efficiency. To tackle the challenges, this research proposes novel online learning models for real-time SPP status prediction. The prediction model is implemented by selecting advanced online learning models to estimate the printing volumes in averages and standard deviations (SDs) considering different pad sizes with different clean ages, directions, printer parameters, etc. The model performances are evaluated in Root Mean Square Error (RMS E), R2, etc. From comparison, the Support Vector Regressor (SVR) shows outstanding prediction performance with R2 values of 92% and 81% for volume averages and SDs. This research emphasizes the potential of using online learning as a preliminary process for effective real-scenario SMT assembly dynamic control.

This research proposes a novel 3D Convolutional Neural Network (CNN) to perform organ tissue segmentation from volumetric 3D medical images. Accurate and efficient segmentation on the 3D medical image of human organ is a critical step towards disease diagnosis. For volumetric 3D medical image segmentation tasks, the effectiveness of conventional 2D CNNs are reduced due to loss of spatial information. To overcome the obstacles, a 3D CNN that implements the convolution and pooling processes in a 3D space is applied as a substitution to the patch division scheme of 2D CNNs. By using 3D CNN, the image becomes scalable in the spatial direction, allowing accurate image detection with different frame sizes. The 3D CNN implements a cube-by-cube scanning strategy, followed by 3D transformation for each cube in terms of convolving and pooling. The segmentation results of 3D CNN is tested on a 3D OCT image dataset of human thyroid. Experimental results demonstrate that the 3D CNN approach obtains outstanding consistency and accuracy.

This research studies a dilated depthwise separable convolution neural network (DSCN) model to identify human tissue types from 3D medical images. 3D medical image classification is a challenging task due to the unpredictable noise and indistinct tissue behaviors of the image content. The objective of this research is to improve typical supervised deep learning model accuracy by using dilated convolution and depthwise separable network approaches on 3D medical image classification tasks. A depthwise separable architecture is used to improve parameter utilization efficiency. Dilated convolutions are applied to systematically aggregate multiscale contextual information and provide a large receptive field with a small number of trainable weights. The performance of the constructed model is tested to perform a multi-class human tissue classification on 3D Optical Coherence Tomography (OCT) images. Experimental results are compared with typical deep learning classification models. The results show that the DSCN model outperforms other models for all the tissue classification tasks. The proposed DSCN model can be a potential approach for the 3D image-based diagnostic tasks in both healthcare and manufacturing field.

The solder paste printing (SPP) is a critical procedure in a surface mount technology (SMT) based assembly line, which is one of the major attributes to the defect of the printed circuit boards (PCBs). The quality of SPP is influenced by multiple factors, such as the squeegee speed, pressure, the stencil separation speed, cleaning frequency, and cleaning profile. During printing, the printer environment is dynamically varying due to the physical change of solder paste, which can result in a dynamic variation of the relationships between the printing results and the influential factors. To reduce the printing defects, it is critical to understand such dynamic relationships. This research focuses on determining the printing performance during printing by implementing a wavelet filtering-based temporal recurrent neural network. To reduce the noise factor in the solder paste inspection (SPI) data, this research applies a three-dimensional dual-tree complex wavelet transformation for low-pass noise filtering and signal reconstruction. A recurrent neural network is utilized to model the performance prediction with low noise interference. Both printing sequence and process setting information are considered in the proposed recurrent network model. The proposed approach is validated using practical dataset and compared with other commonly used data mining approaches. The results show that the proposed wavelet-based multi-dimensional temporal recurrent neural network can effectively predict the printing process performance and can be a high potential approach in reducing the defects and controlling cleaning frequency. The proposed model is expected to advance the current research in the application of smart manufacturing in surface mount technology.