Synchronized Acquisition and Intelligent Cleaning of Multi-Source Monitoring Data in CNC Machining

With the rapid growth of intelligent manufacturing, CNC machining produces large volumes of heterogeneous monitoring data. Closed CNC architectures hinder access to internal signals and real-time synchronization with external sensors, while noise and faults degrade data quality. This paper proposes a synchronized data acquisition and time alignment approach based on the OPC protocol and the LabVIEW platform, using a unified time-base and shared triggering mechanism to achieve synchronization across internal and external sources. Machining state recognition is employed for time-domain segmentation to enhance data quality. To mitigate the impact of anomalies, an intelligent data cleaning method combining autoencoders and angle-based outlier detection (ABOD) is introduced. Experiments on a ball- screw bench show a synchronization error of less than or equal to 0.02 s and more accurate anomaly localization than KNN and LOF, demonstrating the framework's effectiveness for intelligent monitoring and diagnostics.