Predictive modelling of pump noise using multi-linear regression and random-forest models - via optimal data splitting. (11th January 2023)

• Record Type: Journal Article
• Authors: John, Mir Mohsin; Ganiny, Suhail; Hanief, Mohammad
• Is Part Of: International journal of simulation and process modelling. Volume 18:Number 4(2022)
• Page Start: 267
• Page End: 283
• Publication Date: 2023-01-11
• Subjects: axial piston pumps -- APPs -- noise -- modelling -- prediction -- regression -- random forest -- artificial neural network -- ANN -- K-fold cross-validation; Management -- Computer simulation -- Periodicals; Mathematical models -- Periodicals; Operations research -- Periodicals; Simulation methods -- Periodicals; 003.05
• Languages: English
• ISSNs: 1740-2123

Abstract

In this paper, the multi-linear regression and random forest method are used to model and predict the axial piston pump noise. Experimental data is used to model and predict the pump noise as a function of valve seat material, pump speed and pressure. The models are developed using an optimum data proportion, determined using the K-fold cross-validation technique. For comparative analysis, a cascaded neural network is also used for modelling and predicting purposes. Our results reveal that the random forest method is statistically better than the other methods in modelling and predicting the pump noise. Specifically, the mean-squared errors between the three regression models and the neural network model with respect to the experimental data are 10.82, 4.95, 3.97, and 1.26, and the values of the coefficient of determination ( R 2 ) are 0.79, 0.92, 0.93 and 0.96, respectively. The corresponding values for the random forest model are 0.56 and 0.98, respectively.