Development of vibration signal preprocessing and outlier application

Abstract

In most cases, raw data cannot be directly used for data analysis and model training due to common issues such as errors, omissions, duplicates, or outliers, all of which can affect the subsequent construction of models and their accuracy. Moreover, it has been observed during preprocessing that different outlier handling algorithms yield varying results in terms of effectiveness. This essay will specifically optimize vibration signal data collected by sensors. The focus will be on data preprocessing and the analysis of its results, along with outlier handling algorithms, aiming to enhance the accuracy of subsequent modeling. Additionally, a GUI interface will be developed to provide users with convenient data preprocessing functionalities.