Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/7E60BF94-CFB4-416F-8E83-1D4DEDE33094 | ||
| Year | 2022 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Antonios Andreadakis, "Investigation of using machine learning algorithms for predicting the operation of a pumping station", Diploma Work, School of Production Engineering and Management, Technical University of Crete, Chania, Greece, 2022 https://doi.org/10.26233/heallink.tuc.92852 | ||
| Appears in Collections |
In this thesis, the effectiveness of using machine learning algorithms to predict the pressure, flow rate and total energy consumption in a hydraulic network was investigated. Specifically, the network used was a water pumping station with two identical pumps which depresses into a higher level reservoir. Through appropriate manipulation of the valves in the piping network, the pumps can be operated as individual pumps, in series or in parallel. Using the reliable numerical simulation program for hydraulic networks EPANET and for multiple random scenarios of valve position and depression tank level, the values of pressure at the nodes, flow rate in the piping and total energy consumption were calculated. These values were used to train multiple machine learning models - of the regression type - in MATLAB. The model showing the smallest deviation of the root mean square (RMS) was then used to predict the pressure, flow and energy values for 16 predefined scenarios of vane position and reservoir levels. After comparing these predicted values with the corresponding precalculated values from EPANET, the successful feasibility of using machine learning to predict the operation of this pumping station was demonstrated.
Copyright © DIAS 2013
