The aerospace industry is constantly seeking to introduce new technologies that will provide competitive advantages and significant change. Artificial intelligence (AI) and machine learning (ML) approaches in additive manufacturing (AM) offer significant benefits in meeting the industry's needs. Research insights suggest that combining these advanced technologies could streamline existing research and development (R&D) efforts, improve part quality, reduce costs, and enhance overall system efficiency. Despite these benefits, this integration comes with its own challenges, including the need for robust data management systems and the development of reliable and accurate AI algorithms. Furthermore, validating the repeatability and reliability of AI-enabled systems in complex AM aerospace applications raises more questions than concrete answers, necessitating further investigation for widespread adoption.
America Makes, the national additive manufacturing innovation organization, has identified growing industry-wide interest in capturing market needs and further exploring the opportunities and challenges of integrating AI and ML approaches into AM for aerospace applications.
Integrated materials and process qualification
Predicting outcomes in AM can be unclear due to variations in machine, material, and printing parameters. Developing AI and ML models that can be applied across processes can provide significant benefits in producing high-quality parts. However, the complexity and variability of AM processes make it difficult to accurately predict outcomes, especially for high-criticality applications. Rigorous testing, validation, and qualification and certification of required equipment, methods, and personnel are required to ensure the safety of the platform and users. Although significant cost and complexity are involved, developing robust and accurate AI and ML models for AM has demonstrated potential to improve the economics and production efficiency of high-quality AM applications.
As a result, ML represents a huge opportunity for the AM industry to correlate materials with specific parameter sets, achieve consistent and reliable material output, and validate robust processes. Manually analyzing real-time data from large data sets is a time-consuming, complex, and potentially confusing process. However, delegating data analysis to a computer for concurrent processing accelerates industry advancements and delivers statistically validated results.
Recognizing this problem, the laboratory and the National Center for Defense Manufacturing and Machining (NCDMM), under the direction of the Department of Defense (DoD), initiated a $3.2 million project titled, “Demonstration of a New Methodology for Effective AM Process Qualification/Requalification – Delta Qualification.” The overall goal of the project was to demonstrate an AM process that could provide an efficient and cost-effective means to incorporate changes in critical process, post-processing techniques, and material feedstock variables while ensuring validation of qualified AM materials through statistical analysis.
Senvol, a member of America Makes, was awarded a topic within the project focused on leveraging ML to accelerate the delta certification process. New York-based Senvol came up with an innovative approach that uses ML algorithms to calculate statistically-based material property predictions similar to material allowables.
Within the topic area, the Senvol team simultaneously analyzes numerous varying parameters, noting the marginal contribution of each. The study presents a viable alternative to traditional “point solution” methods, opening the door to a more economically feasible and flexible approach and demonstrating that data-driven ML algorithms can significantly reduce the cost of developing material allowances. This approach allows for a deeper understanding of optimal solutions to overcome the qualification and requalification challenges that are slowing the expansion of this innovative industry.
Quality Control and Assurance
AI and ML algorithms are highly dependent on data, and the quality of such data directly impacts the accuracy of the algorithms' predictions and decisions. As AI-powered systems are increasingly being explored to automate various aspects of AM, it is critical to ensure that the end product meets the required quality standards.
Historically, manufacturing quality control and quality assurance practices relied heavily on the ability of human inspectors to scrutinize products for defects and deviations from specifications. This was a tedious, costly process that was prone to errors. Leveraging knowledge gained from increasing R&D efforts, the industry is addressing these technology gaps with a new understanding of the capabilities of AI and ML when combined with AM. This has led to significant improvements in AI automation at various manufacturing stages. Today, AI algorithms help detect potential failures in aircraft parts before they occur by analyzing data collected from sensors. These algorithms can be adjusted in real time, enabling proactive maintenance and preventing costly downtime. This approach also reduces the chances of human error and improves the accuracy of printed parts.
However, the introduction of AI can create new challenges in quality assurance and control, such as the specialized knowledge and skills required to operate and maintain AI-enabled machines. Additionally, the complexity of aerospace parts demands high accuracy and precision that can be difficult to achieve with AM processes.
Despite these challenges, the integration of AI has great potential to improve the quality and efficiency of aerospace part production, provided that appropriate quality assurance and control measures are implemented. Therefore, it is necessary to establish robust quality assurance and control frameworks that can adequately address the challenges that arise. These frameworks should incorporate rigorous testing and validation procedures that ensure the accuracy and reliability of the final product.
Furthermore, human involvement
The complex interplay of AI, ML, and AM must be navigated. Effectively integrating these technologies requires highly skilled professionals with expertise in multiple fields, including computer science, materials science, physics, and aerospace engineering. This challenge can be met by investing in training and education programs that focus on developing interdisciplinary skill sets.
Exploring the possibilities of cutting-edge technology
From a broader perspective, AM offers advantages over traditional manufacturing methods due to its digital nature, allowing for optimized output. Complex parts can be manufactured in a single part, reducing assembly and labor, leading to cost savings. With the help of AI and ML, AM data can be leveraged to improve productivity, yield, and quality control, reducing the need for costly and time-consuming inspection and post-processing.
The current state of affairs is complex, fraught with obstacles, and requires careful consideration to successfully implement AI, ML, and AM in aerospace applications. Overcoming these challenges requires sustained efforts dedicated to developing robust, scalable models that can be applied to a range of AM processes to enable the development of powerful, efficient, and cost-effective manufacturing processes. Regardless of the technical hurdles, the possibilities offered by such advanced technologies justify multidisciplinary collaboration to foster innovative solutions that advance U.S. manufacturing.
