Recent studies in additive manufacturing (AM) monitoring techniques have focussed on the identification of defects using in situ monitoring sensor systems, with the aim of improving overall AM part quality. Much work has focussed on the use of of camera-based monitoring systems; however, limitations such as the slow response rates of the sensors (1-10kHz) and the post-processing requirements of the collected images make it difficult to apply these developmental monitoring methods on production systems in real-time. Furthermore, the replication of results from camera-based monitoring systems (often obtained using deep learning models) in a production environment is limited by the need for specialised hardware with high computational capacity (e.g GPUs). Focussing specifically on laser powder bed fusion ( PBF-L/M ), photodiodes, with fast data collection rates (50–100kHz) and providing data that is relatively easy to process are potentially better suited to real-time monitoring systems. The current study, therefore, focuses on using data collected from photodiodes to identify defects in PBF-L/M builds. A predictive model with real-time potential is proposed that, having been validated on data from computer tomography (CT) images, can be used to locate porosity within layers of PBF-L/M builds.

Recent developments in additive manufacturing (AM) monitoring methods have been introduced to satisfy the quality standards expected in risk averse industries (e.g. aerospace and biomedical implants). However, due to limitations such as slow response rates of the sensors equipped in AM systems and significant image processing requirements, it is difficult to apply these developmental monitoring methods on production systems in real-time. The work herein proposes a real-time predictive model that can be used to indicate defective areas in layers of laser powder bed fusion (LPBF) builds. A second-order auto-regressive (AR) model was trained on a photodiode signal and was used to predict porous regions of the builds. Existing defective areas of the build were first identified using a database of computer tomography (CT) images. Induced defects could provide a poor representation of real defect phenomena. Hence, an L-PBF build with naturally formed defects was examined in this study. By comparing the model predictions with the CT images it was demonstrated that the proposed predictive model is capable of predicting pores, with a minimum diameter of approximately 100 μm, in L-PBF builds with an average true positive rate (TPR) of 88.47%.

While Laser powder bed fusion (L-PBF) machines have greatly improved in recent years, the L-PBF process is still susceptible to several types of defect formation. Among the monitoring methods that have been explored to detect these defects, camera-based systems are the most prevalent. However, using only photodiode measurements to monitor the build process has potential benefits, as photodiode sensors are cost-efficient and typically have a higher sample rate compared to cameras. This study evaluates whether a combination of photodiode sensor measurements, taken during L-PBF builds, can be used to predict measures of the resulting build quality via a purely data-based approach. Using several unsupervised clustering approaches build density is classified with up to 93.54% accuracy using features extracted from three different photodiodes, as well as observations relating to the energy transferred to the material. Subsequently, a supervised learning method (Gaussian Process regression) is used to directly predict build density with a RMS error of 3.65%. The study, therefore, shows the potential for machine-learning algorithms to predict indicators of L-PBF build quality from photodiode build measurements only. This study also shows that, relative to the L-PBF process parameters, photodiode measurements can contribute to additional information regarding L-PBF part quality. Moreover, the work herein describes approaches that are predominantly probabilistic, thus facilitating uncertainty quantification in machine-learnt predictions of L-PBF build quality.

Risk-averse areas such as the medical, aerospace and energy sectors have been somewhat slow towards accepting and applying Additive Manufacturing (AM) in many of their value chains. This is partly because there are still significant uncertainties concerning the quality of AM builds. This paper introduces a machine learning algorithm for the automatic detection of faults in AM products. The approach is semi-supervised in that, during training, it is able to use data from both builds where the resulting components were certified and builds where the quality of the resulting components is unknown. This makes the approach cost efficient, particularly in scenarios where part certification is costly and time consuming. The study specifically analyses Laser Powder-Bed Fusion (L-PBF) builds. Key features are extracted from large sets of photodiode data, obtained during the building of 49 tensile test bars. Ultimate tensile strength (UTS) tests were then used to categorise each bar as ‘faulty’ or ‘acceptable’. Using a variety of approaches (Receiver Operating Characteristic (ROC) curves and 2-fold cross-validation), it is shown that, despite utilising a fraction of the available certification data, the semi-supervised approach can achieve results comparable to a benchmark case where all data points are labelled. The results show that semi-supervised learning is a promising approach for the automatic certification of AM builds that can be implemented at a fraction of the cost currently required.

Risk-averse areas such as the medical, aerospace and energy sectors have been somewhat slow towards accepting and applying Additive Manufacturing (AM) in many of their value chains. This is partly because there are still signicant uncertainties concerning the quality of AM builds. This paper introduces a machine learning algorithm for the automatic detection of faults in AM products. The approach is semi-supervised in that, during training, it is able to use data from both builds where the resulting components were certied and builds where the quality of the resulting components is unknown. This makes the approach cost ecient, particularly in scenarios where part certication is costly and time consuming. The study specically analyses Selective Laser Melting (SLM) builds. Key features are extracted from large sets of photodiode data, obtained during the building of 49 tensile test bars. Ultimate tensile strength (UTS) tests were then used to categorise each bar as `faulty' or `acceptable'. A fully supervised approach identied faulty specimens with a 77% success rate while the semi-supervised approach was able to consistently achieve similar results, despite being trained on a fraction of the available certication data. The results show that semi-supervised learning is a promising approach for the automatic certication of AM builds that can be implemented at a fraction of the cost currently required.

This paper proposes an efficient quadratic Bezier curve based approach for path planning of a quadcopter. It provides a path planning algorithm for the generation of smooth trajectories bounded by dynamic constraints of a quadcopter ensuring less computational load. The trajectories are required to be generated using available coordinates, including starting point and the point of obstacle provided by sensors. The algorithm suggests a method finding the end point and intermediate point in terms of available data. And quadratic Bezier curve is then formed using starting point, ending point and intermediate control point to generate trajectories for the quadcopter. The derived intermediate control point is changed appropriately avoiding collision with the identified obstacle. And it is changed in a manner, providing easy convergence to the original path. Once the path is generated with continuous and bounded curvature, the velocity of the quadcopter along the path is varied continuously and if required varied to its maximum level by ensuring its acceleration is within its acceptable limits. The controlling of acceleration is carried out considering the maximum angle of rotation about its three axes. And projections of planes are used to address the obstacle avoidance situation in a 3 dimensional aspect.