Papers by Author: Er Shun Pan

Abstract: A stencil printing process (SPP) optimization problem is studied in this paper. Due to the limitation that neural network requires a large number of samples for the accurate model fitting, a two-stage SPP optimization method is proposed. The design interval can be reduced with small sample by using neural network. In this reduced design interval , response surface method is adopted to obtain the accurate mathematical SPP model. The concept of confidence level is introduced to make the proposed model robust. An interactive method is used to solve the model. The proposed method is compared with the one-stage optimization method and the results show that the proposed method achieves a better performance on each objective.

Abstract: This paper proposes an effective method to determine the optimal parameters settings of reflow soldering profile and helps reduce the try-and-error time in practical application. Due to the complex nonlinear relationship between reflow thermal profile and process parameters in this problem, BPNN is adopted to establish the model for description of this intricate relationship between inputs and outputs. According to the requirements of the reflow soldering profile, GA is used to calculate the best input parameters with its strong global research ability. Thus, a combined model with BPNN and GA is proposed for parameters optimization and contributes to reduction of decision time of input parameter for reflow soldering profile. In addition, a case study is given to prove the accuracy of the proposed method.