Lasers that Learn: AI-enabled intelligent materials processing

激光被用于非常广泛的制造过程。这部分归因于它们在诸如脉冲长度、脉冲能量、波长和光束大小等参数方面的显著灵活性。然而，这种灵活性是有代价的，即必须为每个制造工艺或客户规格寻找最佳参数集所需的大量时间。工业中的标准做法是机械地收集所有参数组合的激光加工数据，以便找到最佳的参数组合。然而，这一过程既耗时又分散，可能需要几天或几周的时间，因此花费了不必要的时间和金钱。即使确定了最佳参数，在制造过程中，激光功率或光束形状的微小变化也可能导致最终产品质量低于要求的标准，再次耗费时间和金钱。也会存在由于制造过程中的可变性而事先不知道规格的情况。因此，需要一系列的方法来识别制造前的最佳参数，在制造过程中提供实时监控和纠错，并实现过程控制(例如，在任务完成时准确停止激光，或在最后的精加工步骤中改变激光功率)。近年来，机器学习的研究领域取得了一些极其重大的发展，现在人们普遍认为，它是几乎所有制造业发生根本性变化的催化剂。这项提议的目标是开发将机器学习方法整合到英国基于激光的制造业和英国国民健康保险制度所需的技术和人力专业知识。因此，这项提议寻求利用最先进的机器学习技术来解决基于激光的制造和材料加工中的众所周知的问题，从而提高效率、可靠性和精度。这项提议的结果将为英国激光制造业和英国国民健康保险制度节省时间和金钱。这项建议将涵盖应用神经网络对飞秒激光加工进行建模和优化、针对任何客户规格即时识别基于激光的制造参数、自动补偿光纤激光器中的残余空腔效应、实现激光治疗牛皮癣的定向传输，以及通过多传感器数据实时增强激光焊接工艺。

Lasers are used for an extremely wide range of manufacturing processes. This is due, in part, to their significant flexibility with respect to parameters such as pulse length, pulse energy, wavelength, and beam size. However, this flexibility comes at a price, namely the significant amount of time that must be dedicated to finding the optimal set of parameters, for each and every manufacturing process or customer specification. The standard practice in industry is the mechanical collection of laser machining data for all parameter combinations, in order to find the optimal combination of parameters. However, this process is both time-consuming and unfocussed, and it can take days or weeks, hence costing unnecessary time and money. Even when the optimal parameters have been determined, small changes, for example in laser power or beam shape, during manufacturing, can result in a final product quality that is below the required standard, once again costing time and money. There will also be instances where the specification is not known in advance due to variability in the manufacturing process. What is needed, therefore, are a series of methodologies for identifying optimal parameters before manufacturing, for providing real-time monitoring and error correction during manufacturing, and for enabling process-control (for example stopping the laser exactly at task completion, or varying the laser power for the final finishing steps).The research field of machine learning has seen some extremely significant developments in recent years, and it is now widely understood to be a catalyst for a fundamental change across almost all manufacturing industries. The objective of this proposal is to develop the technological and human expertise required for the integration of machine learning approaches into the UK laser-based manufacturing industry and the NHS. This proposal therefore seeks to leverage state-of-the-art machine learning techniques for solving well-known problems in laser-based manufacturing and materials processing, resulting in improvements in efficiency, reliability, and precision. The results of this proposal will lead to time and money savings for both the UK laser-based manufacturing industry and the NHS. This proposal will cover the application of neural networks for modelling and optimising of femtosecond laser machining, instantly identifying laser-based manufacturing parameters for any customer specification, automatically compensating for residual cavity effects in fibre lasers, enabling targeted delivery of laser light for psoriasis treatment, and laser welding process enhancement in real-time via multi-sensor data.