Fibre reinforced polymer composites (FRP) are made of polymer matrix reinforced with fibers. They have much higher strength-to-weight ratio and excellent corrosion resistance compared with steel. However, they are not wildly used because of the lack of understanding on their durability across the long lifespan of civil infrastructures. Unfortunately, durability is an extremely complex problem and which can be affected by many factors, such as fiber type, matrix type, water/moisture, alkali, temperature, load, UV radiation, etc. These factors have a complex coupling effect when combined, making it impossible to build a theoretical model to describe FRP durability in practice. Machine learning becomes the only option for this problem. Machine learning offers several advantages, including the ability to identify hidden patterns in data, make accurate predictions, and perform classification tasks. In this project, I aim to build a high-quality database and develop reliable machine learning algorithms for analysing the durability of FRP composites. They will be eventually integrated into a digital application offering a convenient tool box which can be downloaded for free for engineering design.
Above figure shows that the composite materials are subjected to complicated environmental conditions during their service life and their material properties degrade with time. Unfortunately it is impossible to predict the degradation due to the complex nature of durability. Machine learning can be used for this particular challenge. It learns from the experimental database from the literature and can be developed into a digital tool called "dura tool" for degradation prediction.
