I-Corps: AI-enabled automation intelligence software that can detect micro-anomalies in machine and robotic operations

这个i-Corps项目的更广泛的影响/商业潜力是开发一种启用人工智能(AI)的自动化软件，该软件可以检测机器和机器人操作中的微小异常，并在灾难性设备故障之前精确定位有缺陷的部件。有缺陷的组件会导致计划外停机，这对制造商来说是一个重大挑战。在一个典型的工厂中，工程师可能会花费30%-60%的时间来收集与缺陷部件相关的信息。82%的公司经历过计划外停机，平均每年造成200万美元的损失。然而，70%的制造商仍然不知道设备资产何时需要维护或升级，72%的制造商将计划外停机视为他们的首要任务或高度优先事项。先进的自动化和传感器技术在工业作业中的高渗透率也增加了对系统和及时有效地监测和管理复杂作业的方法的需求。拟议的技术通过向维护和工业工程师提供能力，使昂贵的制造设备故障的平均解决时间(MTTR)接近于零，从而满足了这一需求。此外，该技术通过先进的方法来分析数据并在故障发生后提供见解，从而消除了机器故障的发生。该i-Corps项目基于1)专有的可编程逻辑控制器(PLC)驱动程序的开发，该驱动程序可捕获机器部件的微退化趋势；2)反映物理系统性能和机器之间交互的机器生态系统的数字双胞胎；以及3)自动化智能网络。这项拟议技术独特的机器性能数据集由其PLC驱动程序和传感器反馈创建，可捕获机器的“正常”操作。然后使用这些数据集在系统的不同元素之间创建因果关系图。因果图可加速检测潜在机器故障的根本原因，并预测系统中的潜在风险和弱点。此外，数字双胞胎提供了一个网络物理平台，集成了基于物理的机器部件模型、来自计算机辅助工程(CAE)模型的空间关系，以及从自适应机器学习方法推断的数据驱动的关联。AI平台在分布式系统/边缘分析网络上运行，以确保在本地节点(例如，单个机器及其组件)和系统(例如，生产线)级别上的实时监控和诊断。这三个元素的整合建立了一个可扩展的数字孪生兄弟，利用元素交互来持续发现和细化因果关系，从而快速可靠地检测、识别和精确定位问题。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this I-Corps projects is the development of an Artificial Intelligence (AI)-enabled automation software that detects micro-anomalies in machine and robotic operations and pinpoints defective components before catastrophic equipment failure. Defective components cause unscheduled downtime and are a significant challenge for manufacturers. In a typical factory, an engineer may spend 30-60 percent of their time collecting information related to defective parts. Eighty-two percent of companies have experienced unplanned downtime that costs an average of $2 million annually. However, 70% of manufacturers still lack awareness of when equipment assets are due for maintenance or upgrade, and 72% of manufacturers identify unplanned downtime as their top priority or a high priority. The high penetration level of advanced automation and sensor technologies in industrial operations also has increased demand for methods to effectively monitor and manage complex operations in a systematic and timely manner. The proposed technology addresses this need by delivering capabilities to maintenance and industrial engineers to achieve near zero Mean Time-to-Resolution (MTTR) of costly manufacturing equipment failures. In addition, the technology eliminates the machine failures from occurring by advancing approaches that analyze data and provide insights after failure occurs. This I-Corps project is based on the development of 1) a proprietary Programmable Logic Controller (PLC) driver that captures machine component micro degradation trends; 2) digital twins of the machine ecosystem that mirror the physical system performance and interactions between machines; and 3) an Automation Intelligence network. The proposed technology’s unique machine performance datasets, created from its PLC-driver and sensor feedback, captures the “normal” operation of the machines. These datasets are then used to create causal relationship maps between different elements of the system. Causal maps accelerate the detection of root causes of potential machine failure and anticipate the potential risks and weaknesses within a system. In addition, digital twins provide a cyber-physical platform that integrates physics based models of machine components, spatial relationships from Computer Aided Engineering (CAE) models, and data-driven correlations inferred from self-adapting machine learning methods. The AI platform runs on a distributed system/edge analytics network to ensure real-time monitoring and diagnosis on the local node (e.g., individual machines and its components) and system (e.g., production lines) levels. Integration of these three elements establishes a scalable digital twin that leverages element interactions to continuously discover and refine causal relationships that rapidly and reliably detect, identify, and pinpoint issues.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.