Intelligent Fault Detection for Additive Manufacturing

"Intelligent Fault Detection for Additive manufacturing (IFDAM) is a ground-breaking project to embed machine learning and AI into the laser powder bed fusion (LPBF) process. HiETA Technologies Ltd is a market leader in using LPBF process to produce highly efficient and lightweight thermal management solutions for motorsport, automotive, aerospace and energy applications.Metal AM technologies are seeing a substantial growth and currently rely heavily on post-process inspection methods to determine part build quality. This is costly and time consuming. Failed parts are not identified until significant value has been added to the component as it travels through the production value chain. For example, a defect during the AM build process may not be identified until it has been leak tested. By this point up to 70% additional value has been built into the component. The cost savings which can be achieved through the IFFAM project are considerable through waste reduction in the manufacturing process as HiETA will be able to stop adding value to defective parts.In-process monitoring technology has recently become available but is yet not proving its value to be implemented in the AM process chain. The current challenges are: 1\. Generation of big streams of data at high frequency 2\. Storage, collection and analysis of data, i.e. centralised servers for data storage, absence of data mining and predictive analysis 3\. Absence of control strategies i.e. reactive response (via alerts) rather than corrective response once the defect has been detected.The IFDAM addresses these challenges in the following ways:1\. Adaptive data management i.e. selection of most useful data and development of data dimensionality reduction techniques 2\. Enable mining the relationships between part design, materials, and production processes to predict performance and validate those results against physical test results 3\. Development and validation of corrective/feedback control actions using deep learning algorithms for automated fault detection and correction to reduce the number of post-build inspection and costly certification experiments for the aerospace and energy sectors.

With the support of STFC and NPL, IFDAM will drastically improve the in-process inspection methodologies to allow for in-line quality evaluation of components. Ultimately this will allow HiETA to bottom out root causes of part defects, designing them out for future components massively reducing reject levels and allowing HiETA to use AM to compete on price and quality with more traditional manufacturing processes."