Smart building air condition and mechanical ventilation (ACMV) systems which provide cooling operations suffers from the balance between thermal comfort (TC) and energy consumption (EC). This dissertation proposes a bilayer stochastic multi-objective optimization model that addresses the TC and EC trade-off by maximizing TC in the upper layer. The lower layer can help the ACMV system operate at an optimal frequency to reduce EC. To simultaneously determine the optimal TC and EC, the model is solved using an artificial neural network coupled with a multi-objective whale optimization algorithm. The contribution and novelty of this study pivots on proposing the joint optimal operating condition for TC and component operating frequencies for the energy conservation of the ACMV system. Furthermore, indoor temperature setting method for mutual-improvement of TC and task performance (P) while minimizing EC and the energy-saving potential with respect to the proposed temperature setting method was examined. The preferable temperature range under various combinations of TC and P is recommended. By reducing the preferred upper limit of the temperature range by 3℃ compared to current standards, the algorithm further reduces energy consumption. Additionally, the proposed model effectively captures the complex trade-offs and complementary nature between TC and P with respect to indoor temperature. The proposed indoor temperature setting method saves 9.98 % of energy within the desired temperature range. The present model successfully built and resolved the difficulties of the ACMV system against outdoor and indoor uncertainty in a building. Keywords: Air conditioning system; bi-layer stochastic optimization; model-based control; multi-objective whale optimization algorithm; uncertainties.