AkzoNobel: Revolutionizing Color Prediction with IoT

技术

分析与建模 - 机器学习
分析与建模 - 预测分析

适用行业

教育
设备与机械

适用功能

产品研发
质量保证

用例

预测性维护
时间敏感网络

服务

数据科学服务
测试与认证

挑战

阿克苏诺贝尔是一家荷兰油漆和涂料公司，两个世纪以来一直处于配色领域的前沿。然而，该公司在跟上汽车和室内装饰等行业快速发展的色彩趋势方面面临着挑战。传统的颜色预测方法涉及复杂的数学模型，不再有效或创新。随着新颜色的不断出现，涂料行业面临着巨大的压力，制造商不断寻求新的饰面以获得竞争优势。阿克苏诺贝尔的颜色预测过程涉及破译影响颜色的多个物理元素，非常复杂且耗时。该公司需要创新和适应，以满足客户的现代需求和期望。

关于客户

阿克苏诺贝尔是一家领先的荷兰油漆和涂料公司，业务遍及 150 多个国家，拥有约 34,500 名员工。该公司拥有多乐士和新劲等受欢迎的品牌，并拥有 200 多年的丰富传统。阿克苏诺贝尔始终走在色彩趋势的最前沿，拥有一支专门的科学家团队，致力于调整、重新校准和微调色彩，以满足各行业的最新趋势。该公司以其对创新和超越客户期望的承诺而自豪。阿克苏诺贝尔成功的关键在于其适应和创新的能力，不断寻求新的方法来满足市场不断变化的需求。

解决方案

阿克苏诺贝尔求助于由 Microsoft Azure 提供支持的机器学习技术和人工智能，彻底改变其颜色预测流程。 Azure 机器学习的引入改变了颜色预测的核心流程，并通过新技术对其进行了扩展。该公司现在可以基于深度学习模型进行计算，而不是仅仅依赖物理模型。该技术使阿克苏诺贝尔能够在更短的时间内更准确地创建更多颜色配方。向新技术的过渡是无缝的，实验室技术人员和科学家使用相同的流程和软件工具，但计算更智能，测试轮次更少。在合作伙伴 Machine2Learn 的帮助下，机器学习模型可以轻松部署，而 Azure 的平台即服务 (PaaS) 功能提高了操作简单性，使阿克苏诺贝尔能够根据需要进行扩展或缩减。

运营影响

The implementation of machine learning technology and AI, powered by Microsoft Azure, has revolutionized AkzoNobel's color prediction process. The company can now create more color recipes, more accurately, and in less time. This has not only resulted in significant time savings but also lower costs and a faster time to market. The transition to the new technology was almost seamless, with lab technicians and scientists still using the same process and software tools, but with smarter calculations and fewer rounds of tests. The efficiencies generated by machine learning have also had major benefits for how AkzoNobel serves its customers, particularly in the car industry. The company can now provide the correct recipe of a color that is just introduced in the market more quickly, making car repairs easier and faster.

数量效益

Machine learning technology led to significant time savings of approximately 15-20 percent for the team working on new colors for cars.
The new technology reduced the time to get a car color available for the market from up to two years to just one month.
The time to deploy the company’s first end-to-end environment on Azure was less than two months.