Scholarlink

<p>Copper alloys and copper-based composite materials have received considerable attention in engineering fields due to their excellent electrical conductivity, mechanical properties, and better wear resistance. In the current study, predictive models were developed, trained, and validated to predict the hardness of the material, the wear rate, and the COF based on 200 data sets obtained from previously published literature sources. According to the observations, a low wear rate of 0.002274&#xa0;mm³/m was obtained along with a COF of 0.322 at the optimal condition of 14 wt% reinforcement, applied load of 10&#xa0;N, a sliding speed of 0.5&#xa0;m/s, and a sliding distance of 1500&#xa0;m. Three different ANN training methods, Levenberg-Marquardt, Bayesian Regularization, and Scaled Conjugate Gradient methods, were employed using MATLAB. From the results of the error histogram for Bayesian Regularization, it can be observed that the error is closely concentrated around zero, indicating high accuracy in predicting the wear performance using the ANN-BR model. In addition, using the correlation coefficient (R) of 0.99999 for the training set, 0.99866 for the testing set, and 0.99979 for the overall set. Overall, the results confirm that the predictive models developed on the basis of ANN are very successful in predicting properties of copper-based composite materials with high accuracy.</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

Prediction of wear performance of copper matrix composites using Levenberg-Marquardt, Bayesian Regularization, and scaled conjugate gradient algorithms

  • 期刊论文

  • 发表日期：

    2026年6月25日

查看全文
  • R. Sanjeev Kumar,
  • P. Prem Delphy,
  • S. Gopinath,
  • M. Raj Kumar

摘要

Copper alloys and copper-based composite materials have received considerable attention in engineering fields due to their excellent electrical conductivity, mechanical properties, and better wear resistance. In the current study, predictive models were developed, trained, and validated to predict the hardness of the material, the wear rate, and the COF based on 200 data sets obtained from previously published literature sources. According to the observations, a low wear rate of 0.002274 mm³/m was obtained along with a COF of 0.322 at the optimal condition of 14 wt% reinforcement, applied load of 10 N, a sliding speed of 0.5 m/s, and a sliding distance of 1500 m. Three different ANN training methods, Levenberg-Marquardt, Bayesian Regularization, and Scaled Conjugate Gradient methods, were employed using MATLAB. From the results of the error histogram for Bayesian Regularization, it can be observed that the error is closely concentrated around zero, indicating high accuracy in predicting the wear performance using the ANN-BR model. In addition, using the correlation coefficient (R) of 0.99999 for the training set, 0.99866 for the testing set, and 0.99979 for the overall set. Overall, the results confirm that the predictive models developed on the basis of ANN are very successful in predicting properties of copper-based composite materials with high accuracy.