Hybrid integration of genetic algorithm, artificial neural network and mathematical programming for optimization

In this study, a modified hybrid optimization algorithm (HOA) is presented to address the limitations inherent in traditional optimization techniques, particularly for multi-objective problems. This HOA integrates a genetic algorithm, an artificial neural network and a mathematical programming scheme. The integration aims to overcome the computational inefficiencies and convergence issues present in standalone methods. The effectiveness of the proposed HOA is first validated through two standard test cases. In addition, further results are reported for multi-criteria optimization of the process-induced thermal-mechanical behavior of an anisotropic conductive film (ACF)-based ultra-thin chip-on-film (ACF-UTCOF) package, where the objectives are the minimization of package warpage and adhesive peeling stress, and the maximization of contact stress in the micro-joints. Besides, two different HOA models are proposed, and their results are compared in terms of solution accuracy and efficiency. To explore the residual behaviors of the ACF-UTCOF package during fabrication, a process-dependent modeling technique that incorporates both transient thermal and nonlinear contact finite element analysis and ANSYS element death-birth technique is introduced. The effectiveness of this approach is demonstrated through experimental validation.