Estimation of Optimum Design of a 3-Bar Truss System with Decision Tree Algorithm

Abstract

Truss systems are structures that require care and cost due to the materials and workmanship used. In the design of such systems, the process of optimizing the material volume is necessary for optimum cost. In terms of convenience in application, the production of materials in standard sections also reduces labor costs. In this study, the cross-sectional areas of the bars were optimized to minimize the volume for the design of a 3-bar truss system. Harmony Search Algorithm (HSA), a metaheuristic algorithm inspired by nature, was used in the optimization. A data set was prepared by determining the optimum cross-sectional areas for certain load and stress ranges, and a machine learning prediction model based on the load and stress information with the decision tree classification algorithm was produced. For this purpose, the bar cross-section areas in the data were converted to standard cross-sections and divided into classes. With the produced model, under the desired load and stress values, the bar cross-sectional areas of the system were estimated on a class basis. When the results were examined, it was determined that the prediction model produced with the optimum data was successful at a level of approximately 95% in estimating the bar sections. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.