Boeing EPE (Enterprise Prognostic Environment)
Design and Research Lead
Project Objective
The goal of the EPE project was to build a platform to allow Boeing Engineers and Data scientists to more easily develop, test and deploy prognostic alerts for aircraft to Airlines to ensure maximum uptime of Boeing aircraft. The current process and toolset for creating prognostics is incredibly complex and requires accessing multiple databases, using a variety of offline tools and there are significant technical limitations preventing advanced algorithms and Machine learning models from be leveraged.
Our goal was to significantly streamline this process and allow our internal analytics teams to more rapidly solve customer problems through applied data science.
Design Solution
We developed a strategy and vision for a data science focused platform that allowed engineers and data scientists create alerts using modern data science approaches including machine learning. The EPE platform was designed to help data scientists better understand aircraft sensor data and the systems they relate to, improve communication between engineers and data scientists and allow internal and external customers.
We ran a design sprint focused on the data science process, uncovering key bottlenecks and pain points with current tools and processes and developed prototype design that supported a more agile and efficient process. We ran workshops with Airline and Boeing engineers and mechanics to better understand their needs and pain points to ensure that our system would be able to meet their needs and support existing workflows.
The EPE platform has evolved as we have moved the project along, transitioning from a platform dedicated to driving prognostic alerts into Boeing’s Aircraft Health Management platform into a more holistic tool for data scientists and engineers to rapidly deploy and test alerts that can be incorporated into a number of tools and platforms.
Results
We have been able to design and build a working prototype that allows the deployment of advanced prognostics on full flight data and have verified our approach with customers such as JAL and Air Canada. We are currently iterating on the platform and building out new workflows and functionality to support the end to end data science and engineering workflows. We have proven the ROI of this approach to predictive mainenance, reducing service interuptions due to technical issues.
