Optimization of Extreme Learning Machine Based on Improved Beetle Antennae Search for Slot Die Coating Prediction

In the actual production of slot die coating, the minimum coating thickness and the maximum substrate moving speed could only be judged by production experience, and there was no accurate prediction model due to the nonlinear characteristics of fluid motion. Therefore, building a reasonable and efficient prediction model for slot die coating is now an urgent and challenging task. In this paper, an optimized extreme learning machine (ELM) based on improved beetle antennae search (IBAS) algorithm is proposed for slot die coating prediction. The optimized ELM model can well learn the nonlinear characteristics of the system and make accurate predictions, thus solving the traditional inaccurate empirical judgment. As the prediction accuracy of ELM depends on the selection of weights and biases, the IBAS optimization algorithm is used to quickly search for the optimal value of weights and biases in the ELM network. IBAS algorithm improves the generation mechanism of antennae on the basis of the original algorithm, so that the algorithm can converge quickly. At the same time, the search strategy of the algorithm is improved to avoid falling into the local optimal solution. By predicting the production data of slit coating, the feasibility and effectiveness of IBAS-ELM model are proved.