Tool life prediction of dicing saw based on adaptive golden jackal optimizing GRU

The accuracy of tool life prediction will directly affect the overall efficiency of the dicing saw. Since the tool life of dicing saw is easily affected by different working conditions and the material of the tools themselves, it is difficult to establish an accurate tool life prediction model. Therefore, this paper proposes a prediction model based on adaptive golden jackal optimization (AGJO) gated recurrent unit (GRU) to improve the accuracy of tool life prediction of dicing saw. Specifically, the standard golden jackal optimization (GJO) algorithm suffers from slow convergence in the early stage, low convergence accuracy in the late stage, and easy to fall into local optimization. In this paper, a nonlinear convergence factor and an adaptive weighting factor are introduced to improve the standard GJO algorithm. The AGJO algorithm is then compared with other optimization algorithms such as the GJO algorithm using benchmark functions. Secondly, AGJO is used to optimize the hyperparameters of GRU, and the AGJO-GRU tool life prediction model is constructed. Finally, the effectiveness of the AGJO-GRU prediction model was verified using actual data from the ADT-8230 dicing saw. The experimental results show that the method proposed in this paper can effectively predict the tool life of dicing saws. Compared with the GJO-GRU prediction model, the accuracy of the proposed AGJO-GRU prediction model is improved by 1.96%, and the root mean square error decreased by 27.04%, which has better prediction ability.