Machine Learning for Costing Gas-Turbine Components

Machine Learning (ML), part of Artificial Intelligence, is one of the enabling technologies of Industry 4.0. ML appears to be an effective, affordable, accurate and scalable technique to cost mechanical parts in the early stage of the design process. Despite the cost estimation methods proposed in the literature, their application in specific real industrial contexts (e.g., engineered-to-order products) is minimal. This paper presents an innovative method for developing ML-based parametric cost models. The training data set is generated thanks to an analytical and automatic software tool for cost estimation. The data is subsequently processed using the Cross Industry Standard Process for Data Mining – CRISP-DM method. CRISP-DM is a process model for data science and representation. It provides an overview of the data mining life cycle. Its flexibility and easy customisation allow the creation of a data mining model that fits the goal of this work. The proposed method was employed to develop two cost models (semi-finishing and finishing phases) for components (disks) of a gas turbine. Gradient Boosted Trees turned out to be the best-performing prediction algorithm. Design engineers successfully used the generated cost models while configuring the gas-turbine cross-section.