10105227Explainable AI for inspection of weldsClosedCollaborative R&DInnovate UKThe proposed AI project lies within the remit of three groups of implementation steps for the Government Industrial Strategy: Steer the direction of innovation-led economic growth, Drive the UK's international competitiveness, and Support systematically the development of technology. All three cover our initial target market, the energy sector. Until recently the main consideration driving innovation in this sector was the need for decarbonisation. While not renewable as wind power, nuclear energy is nevertheless the second-largest source of low-carbon electricity in the world behind hydropower. This is why many believe that it is essential for helping countries to reach targets of producing all their energy without releasing greenhouse gases. In his spring budget 2023 speech, the chancellor announced that nuclear power is to be classed as "environmentally sustainable" in the UK's green taxonomy, giving it access to the same investment incentives as renewable energy. As Russia's war in the Ukraine has unfolded, the developed countries have also began reconsidering their industrial strategies to reduce dependence on Russian oil and gas. The UK is considering extending life of its existing nuclear reactors, building new nuclear reactors, including the fleet of mini-reactors and extending the wind turbine fleet. Whatever new strategies are adopted in the UK and elsewhere, there will be an increased need for efficient and reliable inspections of energy infrastructure assets. A validated version of AutoNDE, an explainable AI (Augmented Intelligence) in the form of a coded Decision Tree, which is capable of automatic characterisation of defects in nuclear pressure vessels and wind turbines would be an innovative product reducing the operating cost and increasing safety of industrial installations by allowing for more frequent inspections and wider coverage. Simultaneously it would lead to creation of better paid jobs for more skilled inspection and maintenance engineers: it will free them from repetitive tasks. There is no comparable application software on the market, and for this reason, AutoNDE would require development of a blue ocean strategy to achieve commercialisation. On the one hand, the most promising approach is to locate champions of the product in medium-sized NDT service provider companies, to engage them in data sharing and beta-testing and on the other hand, it is necessary to keep increasing the product functionality. Since engineering structures contain welded components and most common failures are associated with fatigue crack growth, this project is directed at addressing the latter aspect of the commercialisation strategy.