Scholarlink

<p>The present study examines the dry sliding wear behaviour of Nb C103 alloy, a commonly used aerospace and high-temperature niobium-based refractory alloy, under different tribological conditions. Experimentation was carried out by changing the load (98.10–156.96&#xa0;N), sliding velocity (1.68–3.35&#xa0;m/s), and sliding distance (1000–3000&#xa0;m), and observations revealed a significant increase in wear rate with increasing load and distance. The underlying mechanisms of wear changed from mild abrasive to delaminative and oxidative wear, as proven by SEM and EDS analysis. For modelling and predicting wear performance effectively, Artificial Neural Network (ANN) structures were proposed and verified. A fifteen-hidden-neuron two-layer feed-forward ANN recorded a mean relative error of 6.70%, whereas a three-layer ANN with eight and fourteen neurons in two hidden layers, respectively, recorded an enhanced error of 5.41%. Regression and performance plots validated the models consistency in predicting wear behaviour within nonlinear tribological regimes. The ANN method was found to be robust and accurate, providing a cost-efficient method compared to the standard wear testing for high-stakes applications using Nb C103. The first-time integration of comprehensive experimental tribological testing of Nb C‑103 alloy with an advanced, multilayer artificial neural network (ANN) modeling framework tailored for high-temperature aerospace alloys. The study not only demonstrates the predictive capability of ANN in capturing complex nonlinear wear phenomena under varied load, speed, and distance conditions, but also establishes direct correlations between microstructural wear mechanisms and model-predicted responses. The proposed approach offers a scalable methodology to reduce experimental workloads, accelerate alloy evaluation, and enhance data-driven design in mechanical systems operating under extreme tribo‑thermal environments.</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

Artificial neural network-based prediction and microstructural insights of wear in Nb C-103 alloy for aerospace applications

  • 期刊论文

  • 发表日期：

    2025年12月3日

查看全文
  • Aravind Sankeerth Kotha,
  • Pagidi Madhukar,
  • Gurabvaiah Punugupati

摘要

The present study examines the dry sliding wear behaviour of Nb C103 alloy, a commonly used aerospace and high-temperature niobium-based refractory alloy, under different tribological conditions. Experimentation was carried out by changing the load (98.10–156.96 N), sliding velocity (1.68–3.35 m/s), and sliding distance (1000–3000 m), and observations revealed a significant increase in wear rate with increasing load and distance. The underlying mechanisms of wear changed from mild abrasive to delaminative and oxidative wear, as proven by SEM and EDS analysis. For modelling and predicting wear performance effectively, Artificial Neural Network (ANN) structures were proposed and verified. A fifteen-hidden-neuron two-layer feed-forward ANN recorded a mean relative error of 6.70%, whereas a three-layer ANN with eight and fourteen neurons in two hidden layers, respectively, recorded an enhanced error of 5.41%. Regression and performance plots validated the models consistency in predicting wear behaviour within nonlinear tribological regimes. The ANN method was found to be robust and accurate, providing a cost-efficient method compared to the standard wear testing for high-stakes applications using Nb C103. The first-time integration of comprehensive experimental tribological testing of Nb C‑103 alloy with an advanced, multilayer artificial neural network (ANN) modeling framework tailored for high-temperature aerospace alloys. The study not only demonstrates the predictive capability of ANN in capturing complex nonlinear wear phenomena under varied load, speed, and distance conditions, but also establishes direct correlations between microstructural wear mechanisms and model-predicted responses. The proposed approach offers a scalable methodology to reduce experimental workloads, accelerate alloy evaluation, and enhance data-driven design in mechanical systems operating under extreme tribo‑thermal environments.