Sprint 1 | Fundament 1
In the first sprint we did the groundwork for weekly documentation during our project. A timesheet document was created in excel where each teammember can log working hours. We created templates for important documents.
A collaboration with TFMC and USN
In the first sprint we did the groundwork for weekly documentation during our project. A timesheet document was created in excel where each teammember can log working hours. We created templates for important documents.
In this sprint the group decided upon how the retrospective would be done and signed a group contract
In the period of the third sprint the group focused on the first presentation. Considering being the first group out
In this sprint the group visited TFMC's testing facilities in Horten. This sparked a lot of new ideas and increased the understanding of the system.
In this sprint the group started to develop architecture for the ANN. The group also started modeling the marine riser
In this sprint the group completed a lot of the architecture for the ANN. The group also started to run OrcaFlex simulations that will yield data needed later in the process. There were steps taken to investigate the bias of the different sensors as well as how Kalman/complementary filters work and best can be implemented.
In this sprint the group finalized the documentation for the second presentation and started rehersing for it
In this sprint the group handed in the second draft of the documentation and presented their progress. The group recived a lot of possitvie feedback, but also further expectations for the third and final document.
During the ninth sprint the group started seing some results with the Machine Learning. The Convolutional and Recurrent neural networks started to outperform the Feed-forward network. The group also started seing results for their mathematical method as well as the Kalman filtering.
In this sprint the group kept working on the neural networks and the surrounding architecture. The team started to be able to generate simulated sensor data. The group have also implemented a rainflow method to calculate the fatigue over time on a system. Through talks with our external this seemed to yield possitive and promising results.
During the 11th sprint the group started to complete their different technical tasks and the focus shifted more towards documentation. The training of the ANN was still in focus but the surrounding tasks where to be completed. A final architecture for the documentation was created and different responsibilities for the documentation was given.