Scholarlink

<p>At present, many models can achieve high-precision prediction of tool wear value and state, but it is prone to the problems of low generalization and poor stability in the face of multi-working conditions or multi-type tool wear signals. Therefore, this paper proposes a tool wear prediction method based on improved gated recurrent unit (GRU), which obtains the optimal value of the GRU model hyperparameters through Bayesian optimization algorithm (BO), and achieves the effect of improving the prediction performance of the GRU model through the optimal hyperparameters. In order to reflect the prediction effect of the proposed model, this paper extracts features from the signals collected under machining under various working conditions and tool types by time-domain, frequency-domain, and time–frequency domain methods, and then divides the features into training set and test set to construct and verify the model, respectively. The results show that compared with SSAE, PSO-LSSVM, GRU, and RNN, the MAE of the proposed model decreases by 15.2%, 1.41%, 22.9%, and 15.6%, and RMSE decreases by 13.3%, 9.12%, 19.1%, and 14.8%, respectively, which not only reflects high prediction accuracy. It also shows strong stability and generalization.</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

Identification of tool wear status using multi-sensor signals and improved gated recurrent unit

  • 期刊论文

  • 发表日期：

    2025年2月20日

查看全文
  • Zisheng Li,
  • Xiaoping Xiao,
  • Wenjun Zhou,
  • Kai Zhang,
  • Honghao Fu

摘要

At present, many models can achieve high-precision prediction of tool wear value and state, but it is prone to the problems of low generalization and poor stability in the face of multi-working conditions or multi-type tool wear signals. Therefore, this paper proposes a tool wear prediction method based on improved gated recurrent unit (GRU), which obtains the optimal value of the GRU model hyperparameters through Bayesian optimization algorithm (BO), and achieves the effect of improving the prediction performance of the GRU model through the optimal hyperparameters. In order to reflect the prediction effect of the proposed model, this paper extracts features from the signals collected under machining under various working conditions and tool types by time-domain, frequency-domain, and time–frequency domain methods, and then divides the features into training set and test set to construct and verify the model, respectively. The results show that compared with SSAE, PSO-LSSVM, GRU, and RNN, the MAE of the proposed model decreases by 15.2%, 1.41%, 22.9%, and 15.6%, and RMSE decreases by 13.3%, 9.12%, 19.1%, and 14.8%, respectively, which not only reflects high prediction accuracy. It also shows strong stability and generalization.