Scholarlink

<p>Designing compositions to improve yield strength for Ni-based superalloys is a vital operation on aeroengines to improve thrust-weight ratio. A data-driven approach by combining machine learning and theory-physical models is proposed to design high-strength Ni-based superalloys. Using machine learning to select 12 critical features to build predicting model, the design space having 372,763 alloys then is determined. The theory-physical models on matrix stability, precipitate stability and the solvus temperature of γ′ are used to select microstructure. The Gaussian process model is built to predict yield strength, determining Ni-15Co-15Cr-4Mo-4Al-3Ti as designed alloy. Experimental tests verify the designed alloy having the features of microstructure stability, the 1142&#xa0;°C solvus temperature of γ′, and the yield strength of 657 MPa at 850&#xa0;°C. To still further improve yield strength, the strengthening mechanisms on precipitate shearing and coherency strain are discussed in detail. A feasible measure is that scales down the particle diameter of secondary γ′ through increasing the cooling rate of solution treatment. The data-driven superalloy-design way can be used as a typical case of big data technology accelerating alloy design. This work taking data-driven concept has a vital significance on avoiding excessive trial-and-error waste to save costs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Design of Novel High-Strength Ni-Based Superalloys Using a Data-Driven Approach

  • 期刊论文

  • 发表日期：

    2025年6月24日

查看全文
  • Jinli Xie,
  • Qin Hailong,
  • Pei Liu,
  • Songyi Shi,
  • Yixin Cheng,
  • Zhimin Sun,
  • Bin Xu

摘要

Designing compositions to improve yield strength for Ni-based superalloys is a vital operation on aeroengines to improve thrust-weight ratio. A data-driven approach by combining machine learning and theory-physical models is proposed to design high-strength Ni-based superalloys. Using machine learning to select 12 critical features to build predicting model, the design space having 372,763 alloys then is determined. The theory-physical models on matrix stability, precipitate stability and the solvus temperature of γ′ are used to select microstructure. The Gaussian process model is built to predict yield strength, determining Ni-15Co-15Cr-4Mo-4Al-3Ti as designed alloy. Experimental tests verify the designed alloy having the features of microstructure stability, the 1142 °C solvus temperature of γ′, and the yield strength of 657 MPa at 850 °C. To still further improve yield strength, the strengthening mechanisms on precipitate shearing and coherency strain are discussed in detail. A feasible measure is that scales down the particle diameter of secondary γ′ through increasing the cooling rate of solution treatment. The data-driven superalloy-design way can be used as a typical case of big data technology accelerating alloy design. This work taking data-driven concept has a vital significance on avoiding excessive trial-and-error waste to save costs.