The model identifies condition of the building i.e whether it is damaged or undamaged, and if it is damaged then it identifies which joint was damaged specifically.
Project Contributors : https://github.com/aarish790 and https://github.com/sumitesh9
There is a building structure which has 30 joints.
At every joint there is an accelerometer attached, which gives the reading of earthquake sensed at that joint.
Help regarding Project was taken from – Qatar University Grandstand Simulator http://www.structuraldamagedetection.com/benchmark/qugs/
In this project I have used only three joints readings.
merged1 csv file consists of readings when joints are damaged due to the earthquake, Additional column in dataset consists of 2,3 and 4 which is used to distinguish between the three joints.
merged2 csv file consists of readings in undamaged condition and damaged condition both, 0 is used for undamaged condition of the joints and 1 for damaged condition of same joints 2,3,4
Random forest Regressor and ANN (Artificial Neural Network) are used together to achieve the desired result.
For Detailed knowledge of Random forest Regressor refer to : https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html .
For Detailed knowledge of ANN (Artificial Neural Network) refer to : https://en.wikipedia.org/wiki/Artificial_neural_network .
Arbitary values were passed into both to check the result of both the algorithms.
Due to the large size of datasets, they cannot be uploaded to github. FOR DATASETS CONTACT AT mdaarishsid790@gmail.com.
Refer to PPT for help regarding understanding the project. https://github.com/aarish790/EARTHQUAKE-DAMAGE-IDENTIFICATION-IN-BUILDINGS-USING-MACHINE-LEARNING-ALGORITHM/blob/master/EARTHQUAKE%20DAMAGE.pptx