Tool life prediction of dicing saw based on PSO-BP neural network

The quality of the dicing will be impacted if the tool wears out quickly during the dicing operation. If the crew changes the tool in a timely manner, the workpieces’ quality of dicing is guaranteed. Therefore, for actual production, estimating the tool’s remaining usable life (RUL) is crucial. A model was proposed in the research to predict the RUL of dicing tool. The back-propagation neural network (BP) and the particle swarm optimization (PSO) algorithm are combined in the model. The model is also known as the PSO-BP prediction model, where the inertia weight of the PSO method can be changed in a more real-time and dynamic way. After several experiments, comparing the experimental results of the proposed model with two traditional models, it was found that the accuracy of the PSO prediction model improved by 0.664% over the BP prediction model and by 0.661% over the traditional PSO-BP (also called as TPSO-BP) prediction model. This concludes that the proposed prediction model is used to predict the RUL of the tool; the results will be more accurate, so the staff can replace the tool in time to ensure the quality and productivity.