Abstract: Engineering ceramics are widely used in industry for their excellent properties. Given the high precision requirements for grinding such ceramics, the acoustic emission (AE) signal characteristics during the process are analyzed. By applying Copula function correlation analysis, the optimal frequency band and characteristics of the AE signals are identified. Furthermore, based on the random forest (RF) neural network optimized by the whale optimization algorithm (WOA), an intelligent prediction model is constructed to achieve precise prediction of the surface roughness of engineering ceramics, including alumina and zirconia ceramics. The results show that the maximum prediction errors are only 8.32% for partially stabilized zirconia ceramics and 7.71% for alumina ceramics, offering a reliable reference and technical support for real-time intelligent monitoring of grinding quality in engineering ceramics.