Support vector regression with parameter tuning assisted by differential evolution technique: Study on pressure drop of slurry flow in pipeline

Abstract

This paper describes a robust support vector regression (SVR) methodology that offers superior performance for important process engineering problems. The method incorporates hybrid support vector regression and differential evolution technique (SVR-DE) for efficient tuning of SVR meta parameters. The algorithm has been applied for prediction of pressure drop of solid liquid slurry flow. A comparison with selected correlations in the literature showed that the developed SVR correlation noticeably improved prediction of pressure drop over a wide range of operating conditions, physical properties, and pipe diameters.

Modelling the influence of pH, fluoride, and caffeine on the corrosion resistance of TiMo alloys by artificial neural networks developed with differential evolution algorithm

“…DE is inspired from the Darwinian principle of evolution in which only the fittest individual survive to the next generations. In combination with ANNs, DE was applied to solve problems such as: oxygen mass transfer in the presence of oxygen vectors , addition of n‐dodecane in the fermentation systems of bacteria and yeasts , classification of crystalline property of organic compounds , freeze drying of pharmaceuticals , hold‐up of solid‐liquid or pressure drop in slurry flow in pipeline .…”

Application of soft computing techniques to multiphase flow measurement: A review

“…There are few reports on the measurement of phase flowrates or phase fractions of slurry flow using soft computing techniques. However, the parameters for considerations of pipe design, such as pressure drop [67], hold-up [68,69] and critical velocity [70,71] have been predicted with soft computing techniques. It is anticipated that more progress in developing the measurement systems incorporating traditional sensors and soft computing techniques would be achieved for the measurement of gas-liquid, gas-solid and liquid-solid flows in the next few years.…”

The use of differential evolution algorithm for solving chemical engineering problems

“…In addition to the model optimization, an exhaustive search of the SVM parameters was performed; the results indicated that this approach is difficult and does not provide the same level of performance as the SVM-DE combination does. A similar approach was used for studying the pressure drop in slurry flow pipelines (Lahiri and Ghanta 2009a). Based on a set of experimental points, a SVM model for regression with seven inputs, six from the previous study (Lahiri and Ghanta 2008) plus pipe diameter, was determined.…”