Thermal imaging based non-destructive testing for fault detection in cold spray additive manufacturing

Abstract

Cold spray is an advanced additive manufacturing technique where spray particles are accelerated through a supersonic nozzle for deposition onto a substrate. Variations in gas properties (e.g., temperature, pressure, mass flow rate) or nozzle conditions (e.g., wear or clogging) can lead to faults that affect the powder feed rate and deposition quality. These changes are reflected in the substrate’s temperature profile, which are monitored using thermal imaging. This study introduces ThermoAnoNet, a thermal image-based non-destructive testing (NDT) framework for detecting anomalies related to changes in gas dynamics or nozzle conditions. ThermoAnoNet forecasts the substrate’s temperature profile under normal conditions and detects anomalies by identifying deviations that indicate faults, such as increased deposition rates or nozzle clogging. The model, based on deep unsupervised learning, identifies temperature deviations that correlate with faults in the cold spray process. Experimental results show that ThermoAnoNet effectively detects anomalies with 90% accuracy, demonstrating its potential for real-time monitoring, minimizing defects, and enhancing the reliability of the cold spray process.