Deep Learning-Based and Kernel-Based Proxy Models for Nonlinearly Constrained Life-Cycle Production Optimization