Life cycle management of railway tracks including AI-based predictive maintenance using digital product passports

英国政府强调制造业的重要性，称其为“对国家未来的金融稳定至关重要”。2022年，英国133,000家制造商（雇佣270万人）的产值达到6.7万亿英镑，排名世界第8位。然而，计划外停机时间占用了每年生产时间的11%，造成1800亿英镑的损失。初步研究表明，从“反应性”和“预防性”转向“基于人工智能的预测性维护”（PdM），可将这些成本降低30%，故障减少75%，停机时间减少45%。大公司大多在内部开发PdM，因此具有更定制的性质。对于英国中小企业制造业来说，预测性维护还处于起步阶段，早期的调查表明，虽然数字化正在取得明显进展，但维护仍然非常被动，而且通常是手动的。该项目利用尖端的人工智能技术，为中小企业市场构建可扩展的预测性维护解决方案，从而使其成为主流。制造业的未来趋势显然是走向数字化。制造商强调，停机时间的成本越来越高，预测性维护在降低成本和提高生产率方面发挥着至关重要的作用。早期的行业研究和对潜在客户的采访表明，该行业具有成本意识，并对经济上可行的新想法持开放态度。通过准确预测维护需求并在最需要的地方分配资源，我们的项目正在引领工厂效率的新时代，从严格的时间表转向数据驱动的决策，确保运营顺利、经济高效地运行，并减少中断。我们建议为中小企业市场构建一个通用的基于人工智能的预测性维护产品。建立准确的预测性维护技术的一个关键挑战是缺乏故障和故障数据，因为机器或生产线经常在故障前进行维修。预防性维护将机器视为有限环境中的“独立”实体。它没有考虑生命环境中结构属性和物理属性的复杂相互作用和进化机制。为了最好地解决这些问题，我们的愿景是在后期将基于人工智能的预测性维护与数字双胞胎的概念联系起来。我们整体方法的最终目标可能会导致下一代制造工厂，其中健康和安全，库存和设施管理都可以更加自动化，可持续和高效，从而提高产品质量，减少浪费并改善劳动条件，同时保持在英国制造。

The UK Government highlights the importance of its manufacturing industry by labelling it as ‘crucial to the country's future financial stability’. In 2022, the output of 133,000 UK manufacturers (employing 2.7m people) reached £6.7tr, ranking 8th in the world. However, unplanned downtime absorbs 11% of annual production time, costing £180bn. Initial studies indicate that moving from ‘Reactive’ and ‘Preventive’ to ‘AI-based Predictive Maintenance’ (PdM) reduces these costs by 30%, breakdowns by 75%, and downtime by 45%. Large companies develop PdM mostly in-house, thereby taking on a more bespoke nature. For the UK SME manufacturing industry, predictive maintenance is in its infancy, as early investigation indicates that, whilst digitalisation is making clear inroads, maintenance is still very much reactive, and often manual. This project leverages cutting-edge AI technology to build a scalable Predictive Maintenance solution for the SME market, thereby helping it become more mainstream. The future trends in manufacturing are clearly towards digitalisation. Manufacturers are highlighting that the downtime is getting much more costly, and the crucial role of predictive maintenance in reducing costs and boosting productivity. Early industry research and interviews with potential customers indicate that the industry is cost conscious and open towards financially viable new ideas. By predicting maintenance needs accurately and allocating resources where they're most needed, our project is ushering in a new era of factory efficiency, moving away from rigid schedules to data-driven decision-making, ensuring operations run smoothly, cost-effectively, and with fewer disruptions.We propose to build a generic AI-based Predictive Maintenance product for the SME market. A key challenge of building accurate Predictive Maintenance technology is the lack of failure and breakdown data, as machines or manufacturing lines are frequently repaired before failure. Preventive maintenance considers a machine as a “stand-alone” entity in a limited environment. It does not consider the complex interaction and evolution mechanism of structural and physical attributes in a life environment. To best address these issues, our vision is to link AI based Predictive Maintenance with the concept of Digital Twins at a later stage. The ultimate goal of our holistic approach will potentially lead to the Next Generation Manufacturing Plants where health and safety, inventory and facilities management can all be more automated, sustainable, and efficient, thereby improving product quality, reducing waste and enhancing labour conditions, whilst keeping manufacturing in the UK.