When i was Whole Engine Design engineer, one of the main problem facing was the jet engine failed test due to high vibration, till date many additional steps been added to the build process. I tried to correlate with simple linear regression using the measurement data at build but found there is no relationship. Now,I am Certification engineer where the each engine's build record and data converged to the department, with the availble data it revive my curiousity. It was greteful this project idea been supported by the group and warm welcomed by lecturer as actual industry use case project. The PDF format presentation slide and Jupyter codes are attached at this folder.
This is a group project for Data Analysis for Sense Making course was conduction during Covid19. The project is to study the feasibility of how machine learning can reducing the test reject of newly build jet engine due to vibration and how to implementation the solution. Due to Covid-19 the whole course was conducted through Zoom, and the group project was completed through the Whatapps and MS Team meeting with file sharing, the team finally completed in about 6 weeks with a online stream video presentation.
Before we go straight into the data crunching, we started the project planning with a guiding principle based on design think which create solution to solve the high test bed reject rate for the engine production. With the design thinking mindset, we started with looking into the business needs, processes, pain point and how we can help to come out a solution. Nothing go smoothly as expected, differences opinions within team member, results from the machine learning unable to match our prediction etc. As all of us impacted with design thinking concept, we communicate and iterate the solutions with frequent open minded meetings before things go too far and delay the project.
With advancement of machine learning packages, simulation with diffrent models is fast and easy. The crucial part is the data, the foundation of any data analysis must the robust and understanding of how the data been related to features are utmost important leading to sensible solutions.