This paper combines the sequential artificial neural network (NN) machine learning with finite element (FE) modeling to assess the solder joint thermal cycling performance. A glass wafer-level chip-scale package (G-WLCSP) is used for this study. This paper investigates the network structure that can achieve prediction capability both inside and outside the design domain with the minimal required training dataset. First, a detailed FE model for G-WLCSP is developed to obtain the accumulated plastic strain per cycle for thermal-cycling loading. Three critical input parameters are defined to generate a dataset based on finite element analysis. Then, applying the supervised machine learning procedure, both the recurrent neural network (RNN) and the gate-network long short-term memory (LSTM) architecture are used to train the obtained dataset. The network complexity of the sequential NN model is carefully controlled to prevent numerical overfitting. Among the total 81 FE generated data pairs, only 27 data pairs have been applied to the sequential NN learning. These 27 data pairs are carefully selected to evenly distributed among the design domain. The average error norms after the learning are 1.213·10-4 and 1.190·10-4of RNN and LSTM, respectively. The prediction capability of the well-trained sequential NN model against the rest 54 data pairs has been tested and a similar scale has been obtained. Furthermore, the prediction capability is tested against the parameters outside the design domain. Approximately one order average error norm increased for both the well-trained RNN and LSTM model.

The performance and reliability of the light-emitting diode (LED) system significantly depend on the thermal–mechanical loading-enhanced multiple degradation mechanisms and their interactions. The complexity of the LED system restricts the theoretical understanding of the root causes of the luminous fluctuation or the establishment of the direct correlation between the thermal aging loading and the luminous outputs. This paper applies the deep machine learning techniques and develops a gated network with the two-step learning algorithm to build the empirical relationship between the design parameters and the thermal aging loading and the luminous output of LED products. The flexibility of the proposed method will be demonstrated by integrating it with different neural network architectures. The proposed gated network concept has been validated in both multiple LED chip packaging and LED luminaire under thermal aging loading. The validation of the luminous data of multiple LED chip packaging shows that the maximum differences of the correlated color temperature (CCT) and color coordinate are 2.6% and 1.0%, respectively. Moreover, the machine learning results of the LED luminaire exhibit that the differences of lumen depreciation, CCT and color coordinate are 1.6%, 1.9% and 1.1%, after 2160 h of thermal aging.

Solder joint fatigue is one of the critical failure modes in ball-grid array packaging. Because the reliability test is time-consuming and geometrical/material nonlinearities are required for the physics-driven model, the AI-assisted simulation framework is developed to establish the risk estimation capability against the design and process parameters. Due to the time-dependent and nonlinear characteristics of the solder joint fatigue failure, this research follows the AI-assisted simulation framework and builds the non-sequential artificial neural network (ANN) and sequential recurrent neural network (RNN) architectures. Both are investigated to understand their capability of abstracting the time-dependent solder joint fatigue knowledge from the dataset. Moreover, this research applies the genetic algorithm (GA) optimization to decrease the influence of the initial guessings, including the weightings and bias of the neural network architectures. In this research, two GA optimizers are developed, including the “back-to-original” and “progressing” ones. Moreover, we apply the principal component analysis (PCA) to the GA optimization results to obtain the PCA gene. The prediction error of all neural network models is within 0.15% under GA optimized PCA gene. There is no clear statistical evidence that RNN is better than ANN in the wafer level chip-scaled packaging (WLCSP) solder joint reliability risk estimation when the GA optimizer is applied to minimize the impact of the initial AI model. Hence, a stable optimization with a broad design domain can be realized by an ANN model with a faster training speed than RNN, even though solder fatigue is a time-dependent mechanical behavior.

The design of the connectors and implant abutments could affect the stress distribution of the tooth-implant supported prosthesis (TISP) entire system after loading. Therefore, the purpose of this study was to investigate the stress distribution of the TISP in different connectors and different implant abutments after loading. The TISP design used in this study was divided into six models. R1, R2 and R3 represented the tooth and the one-piece, two-piece and three-piece abutment implant system connected by a rigid connector, respectively, while NR1, NR2 and NR3 were the corresponding tooth-abutment implant systems connected by a non-rigid connector. A vertical occlusal load of 50 N was applied at a right angle on the 6 occlusal points of the occlusal surface. As a result, regarding the maximum average stress distribution, R1 and NR1 appeared on the implant fixture, and the other four models were on the implant abutment. On the other hand, regardless of the abutment implant system, the maximum von Mises stress generated by the rigid connector was greater than the corresponding non-rigid connector in the cortical bone around implant. In addition, the three-piece abutment implant system had lower von Mises stress than the one-piece and two-piece implant systems in the cortical bone. It is concluded that by adding a flexible non-rigid connector and three-piece abutment device design to TISP, the occlusal load of the implant was dispersed, and the stress could be gradually introduced into the relatively strong implant abutment.

Bridge stability under loading was influenced by bridge span with the connector and implant abutment design. Thus, the purpose of this study was to evaluate the effects of rigid and non-rigid connector designs and pontic connections of different abutment systems in the tooth-implant supported prosthesis (TISP) at different span distances on the biomechanical stress distribution of the overall system components. For comparative analysis, rigid and non-rigid bridge connections were fitted with three implant abutment systems (one-piece, two-piece and three-piece), and five implant-to-natural tooth distance configurations (12 mm, 14 mm, 16 mm, 18 mm, and 20 mm) were provided. The maximum stress between TISP components occurred at the distal side of crown margin of cement1 in rigid connector with one-piece group and the bottom of the crown3 in non-rigid connector with one-piece group, while the other groups were more concentrated at the junction between the mesial side of the implant collar and the abutment. In addition, neither the rigid nor non-rigid connector model showed that stress distribution increased proportionally with the bridge span distance. It was clinically recommended that if the implant with a shorter bridge distance of 12 mm from the natural tooth, the rigid connection of the three-piece abutment can be used as the TISP design. If the bridge distance was 18 mm longer, the non-rigid connection of the three-piece abutment could maintain the physiological movement of the natural tooth and avoid the excessive stress on the bone crest around the implant.

Equation-Informed Neural Networks (EINNs) are developed as an efficient method for extracting the coefficients of constitutive equations. Subsequently, numerical Bayesian Inference (BI) iterations were applied to estimate the distribution of these coefficients, thereby further refining them. We could generate coefficients optimally aligned with the targeted application scenario by carefully adjusting pre-processing mapping parameters and identifying dataset preferences. Leveraging graphical representation techniques, the EINNs formulation is implemented in temperature- and strain-rate-dependent hyperbolic Garofalo, Anand, and Chaboche constitutive models to extract the corresponding coefficients for lead-free SAC305 solder material. The performance of the EINNs-based extracted coefficients, obtained from experimental results of SAC305 solder material, is comparable to existing studies. The methodology offers the dual advantage of providing the coefficients’ value and distribution against the training dataset.