Machine Learning-Driven Topology Optimization Approaches

Chuo Engineering is engaged in design and development work primarily in the aerospace field, but it also does work in the automotive, industrial machinery, telecommunications, home appliances, medical equipment, and software sectors. To address challenges posed by their use of traditional topology optimization, Chuo Engineering turned to machine learning not only for lightweight analysis, but also for multi-disciplinary applications. Using HyperStudy and OptiStruct they were able to eliminate unnecessary analyses to find the best results. Their engineering team used Altair Knowledge Studio to analyze solutions with fewer intermediate density factors among the optimal results obtained in these processes to derive optimal constraints with fewer intermediate density factors through cluster analysis. This helped them reduce the number of times they needed to perform topology optimization analysis of existing metal parts by 93%. In addition, it reduced weight by 54%, which was the objective of the topology optimization analysis.