AI6S – AI lean six-sigma process optimisation for energy efficiency and waste reduction in foundation industries

AI6S将为基础工业（FI）的工艺优化开发一种新颖的工具包，从而提高工艺效率，从而降低能耗，减少浪费，提高满足具有挑战性（和商业吸引力）规范和缩短周转时间的能力。当前的精益六西格玛方法通过在过程迭代中遵循高度手动的程序进行持续改进，从而提高效率。AI6S将使公司实现“正确的第一次”生产能力发生重大变化，以满足具有挑战性的规格和严格的质量标准，同时通过更高的产量，更高效的流程和更快的吞吐量减少能源消耗，浪费和成本，英国每年可能节省21,800,111公斤二氧化碳当量（二氧化碳当量）/ 5800万英镑，全球每年可能节省6.75亿公斤二氧化碳当量/ 18亿英镑。这些好处将通过使用专门针对基础行业（fi）需求量身定制的机器学习优化方法来实现，并将其集成到精益六西格玛框架中。该系统将自动化并大规模加速流程改进和历史数据的重用，以优化流程条件。我们方法的新颖方面包括使用特殊的两阶段优化过程，即使涉及高能热过程的复杂过程也能提供快速的全局优化，并使用数据合成来实现更快，更准确的模型训练。这种新颖的技术方法将被集成到一个精益六西格玛框架中，以供基础行业内的从业人员快速采用。这些算法将在一个软件平台上实现，以方便使用，并与其他质量和企业软件工具集成。该工具包的有效性将通过金属锻造和玻璃生产的两个用例来展示，为两个基础行业公司提供工艺性能的直接收益。

AI6S will develop a novel toolkit for process optimisation in foundation industries (FI), enabling process efficiency improvements resulting in reduced energy consumption and reduced waste and improved ability to meet challenging (and commercially attractive) specifications and short turnaround times. Current lean six-sigma methodologies provide efficiency gains through continuous improvement following highly manual procedures over process iterations. AI6S will make a step change in the ability of companies to achieve "right first time" production output to challenging specifications and stringent quality criteria whilst reducing energy consumption, waste and cost through higher yields, more efficient processes and faster throughput with potential for annual savings 21,800,111 kgCO2e (carbon dioxide equivalent)/ £58M for the UK and 675million kgCO2e/ £1,800million globally.These benefits will be achieved through the use of a machine-learning optimisation approach tailored, for the first time, specifically to the needs of foundation industries (FIs) and integrated within a lean six-sigma framework. The system will automate and massively speed up process improvement and reuse of historical data to optimise process conditions. Novel aspects of our approach include the use of a special two-stage optimisation process which provides rapid global optimisation even for complex processes involved in high energy thermal processes and the use of data synthesis to achieve faster and more accurate model training.This novel technological approach will be integrated into a lean six-sigma framework for rapid adoption by practitioners within the foundation industries. The algorithms will be implemented in a software platform for ease of use and integration with other quality and enterprise software tools. The effectiveness of the toolkit will be demonstrated trough two use cases in metal forging and glass production, providing the two foundation industry companies with direct gains in process performance.