Evolutionary Virtual Expert System

进化虚拟专家系统

• 批准号: EP/R029741/1
• 负责人: Yu Zhang
• 金额: $ 12.28万
• 依托单位: University of Lincoln
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2018
• 资助国家: 英国
• 起止时间: 2018 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FR029741%2F1
• 关键词: Evolutionary; Virtual; Expert; System

UK industries are facing a growing problem - a lack of experts! Multiple sectors of the UK's economy, especially in Engineering, are increasingly dependent on older workers, leaving employers exposed to a massive need for skilled staff when they retire. While the UK attempts to provide more quality vocational training to young people so they can replace skilled older workers when they retire, there remains years of knowledge gap to be filled. Hence, a technological solution becomes increasingly attractive - i.e. assisting humans with "Virtual Expert" (VE) systems and complementing them while they acquire experience. Many UK companies in industry have a range of automation and digitalisation challenges, such as automatic remote condition monitoring tools and engine test automation, which this project seeks to address. The main concept behind this new project is to build and train an Evolutionary Virtual Expert System (EVES) to assist current and future industrial fault diagnostic engineers. These "virtual apprentices" (diagnostic agents, including knowledge-based rules, signal processing algorithms and model-based approaches) will be trained by human experts, through coaching, examining and refining processes. After a number of subject matter tests, the successful "virtual apprentices" are promoted to become VEs and their weightings (rankings) will be updated using a genetic algorithm. Over generations of evolution, EVES will be able to find a suitable population of VEs (rules/algorithms/models), and produce a heuristically best decision through a weighted voting process, with reasoning mechanisms and possible solutions made transparent to users. EVES integrates the strengths of precision, learning ability, adaptability and knowledge representation from all the VEs that conform to the population, aiming to provide an automated and digitalised fault diagnostic system, to match or possibly outperform human experts working without such support.The EVES project will have a big impact on areas of industrial application. This proposal is timely, as the proportion of experts in UK industries are getting older, while at the same time more modern technologies involve longer learning curves for young people. To be ready for the industries of the future, these VEs, when fully trained, will provide critical support for existing experts, and also act as good trainers for the younger workers. As the future generation is based on high technologies, good virtual assistants and virtual trainers will become increasingly important. The proposal is important, as the structure of EVES is widely applicable to all industrial sectors, for example, from fault diagnostics of machines and plants, to remote condition monitoring for railway applications, agriculture precision, water quality monitoring, and even to diagnostics for human health.

英国工业正面临一个日益严重的问题-缺乏专家！英国经济的多个部门，特别是工程部门，越来越依赖老年工人，这使得雇主在退休时面临对熟练员工的大量需求。虽然英国试图为年轻人提供更高质量的职业培训，以便他们在退休时能够取代熟练的老年工人，但仍有多年的知识差距需要填补。因此，技术解决方案变得越来越有吸引力-即用“虚拟专家”（VE）系统帮助人类，并在他们获得经验时对其进行补充。许多英国工业公司面临着一系列自动化和数字化挑战，例如自动远程状态监测工具和发动机测试自动化，该项目旨在解决这些问题。这个新项目背后的主要概念是建立和训练一个进化的虚拟专家系统（EVES），以帮助当前和未来的工业故障诊断工程师。这些“虚拟学徒”（诊断代理，包括基于知识的规则，信号处理算法和基于模型的方法）将由人类专家通过辅导，检查和改进过程进行培训。经过一系列主题测试后，成功的“虚拟学徒”将晋升为虚拟学徒，他们的权重（排名）将使用遗传算法进行更新。经过几代人的进化，EVES将能够找到合适的VE（规则/算法/模型）群体，并通过加权投票过程产生最佳决策，推理机制和可能的解决方案对用户透明。EVES集成了所有符合人群的虚拟环境在精度、学习能力、适应性和知识表达方面的优势，旨在提供一个自动化和数字化的故障诊断系统，以匹配或可能超越人类专家在没有这些支持的情况下工作。EVES项目将对工业应用领域产生重大影响。这一建议是及时的，因为英国各行业的专家比例越来越老，而与此同时，更现代的技术需要年轻人更长的学习曲线。为了为未来的行业做好准备，这些经过全面培训的虚拟企业将为现有专家提供关键支持，并为年轻工人提供良好的培训。随着下一代以高科技为基础，良好的虚拟助手和虚拟培训师将变得越来越重要。该建议非常重要，因为EVES的结构广泛适用于所有工业部门，例如，从机器和工厂的故障诊断到铁路应用的远程状态监测，农业精度，水质监测，甚至到人类健康诊断。

项目成果

Multi-region System Modelling by using Genetic Programming to Extract Rule Consequent Functions in a TSK Fuzzy System

使用遗传编程提取 TSK 模糊系统中规则结果函数的多区域系统建模

• DOI: 10.1109/icarm.2019.8834163
• 发表时间: 2019
• 作者: Zhang Y

Entropy Measures in Machine Fault Diagnosis: Insights and Applications

• DOI: 10.1109/tim.2020.2981220
• 发表时间: 2020-06-01
• 期刊: IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
• 影响因子: 5.6
• 作者: Huo, Zhiqiang;Martinez-Garcia, Miguel;Shu, Lei
• 通讯作者: Shu, Lei

Deep Recurrent Entropy Adaptive Model for System Reliability Monitoring

• DOI: 10.1109/tii.2020.3007152
• 发表时间: 2021-02-01
• 期刊: IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
• 影响因子: 12.3
• 作者: Martinez-Garcia, Miguel;Zhang, Yu;Zhang, Yu-Dong
• 通讯作者: Zhang, Yu-Dong