2020/12/08 by Opeoluwa Owoyele, Pinaki Pal, Owoyele, Opeoluwa +1
Computer Science · Engineering · #Advanced Multi-Objective Optimization Algorithms #FOS: Computer and information sciences #Machine Learning (cs.LG) #Refrigeration and Air Conditioning Technologies
paper · pdf · doi:10.48550/arxiv.2012.04649
openalex publication_date 2020/12/08 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
A novel design optimization approach (ActivO) that employs an ensemble of\nmachine learning algorithms is presented. The proposed approach is a\nsurrogate-based scheme, where the predictions of a weak leaner and a strong\nlearner are utilized within an active learning loop. The weak learner is used\nto identify promising regions within the design space to explore, while the\nstrong learner is used to determine the exact location of the optimum within\npromising regions. For each design iteration, exploration is done by randomly\nselecting evaluation points within regions where the weak learner-predicted\nfitness is high. The global optimum obtained by using the strong learner as a\nsurrogate is also evaluated to enable rapid convergence once the most promising\nregion has been identified. First, the performance of ActivO was compared\nagainst five other optimizers on a cosine mixture function with 25 local optima\nand one global optimum. In the second problem, the objective was to minimize\nindicated specific fuel consumption of a compression-ignition internal\ncombustion (IC) engine while adhering to desired constraints associated with\nin-cylinder pressure and emissions. Here, the efficacy of the proposed approach\nis compared to that of a genetic algorithm, which is widely used within the\ninternal combustion engine community for engine optimization, showing that\nActivO reduces the number of function evaluations needed to reach the global\noptimum, and thereby time-to-design by 80%. Furthermore, the optimization of\nengine design parameters leads to savings of around 1.9% in energy consumption,\nwhile maintaining operability and acceptable pollutant emissions.\n