MIE.37 – Model to Predict the Impact Force from Volume Distribution Data for Spray Droplets
Team Members Heading link
- Eric Anderson
- Eli Gonzalez
- Gitesh Jain
- Joey Malesich
- Kashyap Thakor
Project Description Heading link
Spraying Systems Co. needed to develop a modeling system to bridge the results from the Laser Sheet Imaging data, and High Impact Tester data. The project focuses on developing the model in such a way as to find correlations between the force data from the High Impact Tester, and the intensity data from the Laser Sheet Imaging instrumentation. Upon analysis of the data, the team found that both these data collection methods based their results largely on the distribution and the intensity of the spray drops. When both methods were tested under the same conditions, the results showed a correlation between their profiles. To test this and find the mathematical model for the link between the data, the team developed an algorithm to find the equations to fit the centerline data for each of the results. This algorithm was adjusted and polished over a range of different combinations to maintain a consistent form for the varied conditions consisting of nozzle type, pressure, and distance. Using these parameters, a variety of machine learning approaches were tested to find the most suitable solution for this model. It was found that eXtreme Gradient Boosting was the most accommodating for this project. This algorithm can be used to develop the mathematical model and can be continuously improved by hyperparameter tuning and additional data.