Implementation of ML algorithms that help identify extent of damage in the structure.
Project Contributors - Sumitesh Naithani ([email protected]) Md Aarish Siddiqui ([email protected])
Introduction - This is ML based project trained on dataset of structure (concrete - building) pertaining to 30 joints. In this project we took dataset of only 3 joints. At every joint there is accelorometer attatched to help fetch the reading of the earthquake sensed at that point.
Description of the dataset - merged1.csv - consists of accelerometer readings when the joints are damaged due to earthquake and additional columns are added to distinguish between 3 joints.
merged2.csv - consists of accelerometer readings in both damaged and undamaged condition. Binary classification is used. 0 for undamaged condition and 1 for damaged condition.
Algorithms & Concepts used - 1) Random Forest Algorithm https://en.wikipedia.org/wiki/Random_forest 2) Artificial Neural Networks https://en.wikipedia.org/wiki/Artificial_neural_network
Model is also capable of testing arbitrary accelerometer readings as an input.
Dataset is not uploaded for copyright issues. For further queries check the earthquakedamage.pptx file. https://github.com/sumitesh9/Structural-Damage-Detection/blob/master/EARTHQUAKE%20DAMAGE.pptx