Machine Learning for condition monitoring of production equipment

与传统的“检查和更换”维护计划相比，我的项目将帮助制造商延长生产设备的使用寿命，并使其使用时间更长。通过持续监控设备状况并实时智能洞察持续性能，制造商将能够在需要更改生产计划或减少操作限制以增加被监控设备的使用寿命时，就何时更换设备上的组件做出明智的决策。我们的解决方案的基础是一个专有的人工智能和机器学习软件系统，该系统将使用从传感器（连接到生产设备）获得的数据来发现行为的变化，并在故障发生之前突出潜在的故障点。传感器（如振动、电压、热量、气流、空气微粒等）将捕获数据，并将其显示在数字仪表板上，显示当前的运行性能以及已设置的任何边界（如操作或立法限制），并用标签突出显示假定出现问题的地方——如轴承故障、过滤器堵塞或轴不对中。这个解决方案本身并不独特，但创新之处在于它可以在没有互联网连接的情况下运行，并且完全包含在制造设施中。这种性质的其他系统依赖云计算对感知数据进行人工智能分析，其中数据通过互联网发送到远程计算机，并以相同的方式返回分析。然而，在一些设施中，例如与国防或国家安全有关的设施，不允许任何数据离开设施，因此将使基于云的解决方案失效。相比之下，我们的方法是在提供的服务器中处理所有的人工智能处理，该服务器位于办公场所内，不需要将数据发送到其他地方进行处理。使其免于传输延迟、外部安全漏洞，并可在根本没有互联网连接的地方运行。

My project will help manufacturers increase the lifetime of their production equipment and keep it in use longer - when compared with traditional 'inspect and replace' maintenance schedules.By continually monitoring the condition of equipment and giving realtime intelligent insights into ongoing performance, manufacturers will be able to make informed decisions on when to replace components on their equipment, if there is a need to change production schedules or to reduce operating limits to increase the lifetime of the equipment being monitored.Underpinning our solution is a proprietry Artificial Intelligence and Machine Learning software system that will use data obtained from sensors (attached to production equipment) to spot changes in behaviour, and highlight potential points of failure before they occur. The sensors (such as vibration, voltage, heat, airflow, airborne particulates etc) will capture data that will be represented on a digital dashboard and show the current operating performance along with any boundaries that have been set (such as operational or legislative limits) and highlight with labels what is assumed to be going wrong - such as a bearing failure, filter blockage or shaft misalignment.This solution in itself is not unique, but what is innovative is that it can operate without an internet connection and is fully contained within the manufacturing facility. Other systems of this nature rely on cloud computing for the AI analysis of the sensory data, where data is sent across the internet to a remote computer and the analysis returned the same way. However, in some installations, such as those concerned with defence or national security, no data will be allowed to leave the facility and will therefore incapacitate a cloud based solution.Our approach, by contrast, handles all of the Artificial Intelligence processing from within a supplied server that is placed within the premises and has no need to send data elsewhere for processing. Making it free from transmission lag, external security breaches and can be operated in locations where there is no internet connection at all.