Prediction of surface roughness based on fused features and ISSA-DBN in milling of die steel P20

Abstract

The roughness of the part surface is one of the most crucial standards for evaluating machining quality due to its relationship with service performance. For a preferable comprehension of the evolution of surface roughness, this study proposes a novel surface roughness prediction model on the basis of the unity of fuse d signal features and deep learning architecture. The force and vibration signals produced in the milling of P20 die steel are collected, and time and frequency domain feature from the acquired signals are extracted by variational modal decomposition. The GA-MI algorithm is taken to select the signal features that are relevant to the surface roughness of the workpiece. The optimal feature subset is analyzed and used as the input of the prediction model. DBN is adopted to estimate the surface roughness and the model parameters are optimized by ISSA. (Reviewer 1, Q1) The separate force, vibration and fusion signal information are brought into the DBN model and the ISSA-DBN model for the prediction of surface roughness, and the results show that the accuracy of the roughness prediction is as follows, respectively DBN: 78.1%, 68.8% and 84.4%, and ISSA-DBN: 93.8%, 87.5% and 100%.

Predicting Surface Roughness in Turning Complex-Structured Workpieces Using Vibration-Signal-Based Gaussian Process Regression

“…Recent advancements, including the application of AI for predicting surface roughness in additively manufactured components, as demonstrated by Temesgen Batu et al [35], the utilization of causality-driven feature selection to enhance deep-learning-based models in milling machines by Hyeon-Uk Lee et al [36], and the investigation of novel parameters in grinding processes by Mohammadjafar Hadad et al [37], collectively suggest that innovative methodologies can markedly improve predictive accuracy. The enhanced prediction of surface roughness in titanium alloy during abrasive belt grinding, achieved through an advanced Radial Basis Function (RBF) neural network by Kun Shan et al [38], and the high precision attained by integrating hybrid features with an Improved Sparrow Search Algorithm-Deep Belief Network (ISSA-DBN) for milling die steel P20, as reported by Miaoxian Guo et al [39], further highlight the efficacy of these cutting-edge approaches.…”

Prediction of surface roughness using deep learning and data augmentation

“…To provide candidate features with enough feature selection information to build an accurate workpiece surface roughness monitoring model, this paper extracted 17 time-domain features, 5 frequency-domain features, and 4 time-frequency-domain features from milling force signals and vibration signals in each direction (Guo et al, 2023), resulting in a total of 26 3 6 ¼ 156 features (Table 1). There are four other MATLAB properties of spectral kurtosis in addition to these: the mean of the spectral kurtosis, the standard deviation of the spectral kurtosis, the skewness of the spectral kurtosis and the magnitude of the spectral kurtosis.…”