Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method
Different method of modelling presented in this paper on Shre Drag Tilt Velocimeter non-linear data. The idea of different non-linear modelling method is to know which makes more possible to describe more accurate on interacting effects between velocities and tilt angle when compared among modellers...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Published: |
Universiti Teknikal Malaysia Melaka
2016
|
| Subjects: |
| _version_ | 1796861806124204032 |
|---|---|
| author | Muharram, I. A. M. Ismail, Z. H. |
| author_facet | Muharram, I. A. M. Ismail, Z. H. |
| author_sort | Muharram, I. A. M. |
| collection | ePrints |
| description | Different method of modelling presented in this paper on Shre Drag Tilt Velocimeter non-linear data. The idea of different non-linear modelling method is to know which makes more possible to describe more accurate on interacting effects between velocities and tilt angle when compared among modellers. The models, which were used are static analytic approximation model, curvilinear bivariate regression model, Gompertz the classical growth model and Artificial Neural Network (ANN) model. Accuracy of the models was determined by mean square error (MSE), mean absolute deviation (MAD), bias and R Square. The datasets gathered from an experiment of Shre DTV at flume were divided into training data and testing data for the purpose of developing and validating all type of models. The difference between the model and the observed value become the forecasting error measurements. For the training data, the lowest MSE, RMSE and better R Square were noted for the Gompertz model. But ANN generalized better on testing data by obtaining lowest MSE, RMSE and higher R Square among others. ANN generalization result is 88.60%, Gompertz is 54.89%, curvilinear is 69.28% and static analytic is -1.29%. Lower bias was also for the neural network test data. As demonstrated by the bias values, only curvilinear model presenting overestimation model while other models produce little or no overestimation of the observed tilt response. Interpretations of the parameters estimation on Gompertz model have been attempted previously. However, focusing on the ability of Shre DTV to predict responses may be more practical than the relevance of parameter estimates. |
| first_indexed | 2024-03-05T20:01:59Z |
| format | Article |
| id | utm.eprints-71669 |
| institution | Universiti Teknologi Malaysia - ePrints |
| last_indexed | 2024-03-05T20:01:59Z |
| publishDate | 2016 |
| publisher | Universiti Teknikal Malaysia Melaka |
| record_format | dspace |
| spelling | utm.eprints-716692017-11-22T12:07:38Z http://eprints.utm.my/71669/ Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method Muharram, I. A. M. Ismail, Z. H. TK Electrical engineering. Electronics Nuclear engineering Different method of modelling presented in this paper on Shre Drag Tilt Velocimeter non-linear data. The idea of different non-linear modelling method is to know which makes more possible to describe more accurate on interacting effects between velocities and tilt angle when compared among modellers. The models, which were used are static analytic approximation model, curvilinear bivariate regression model, Gompertz the classical growth model and Artificial Neural Network (ANN) model. Accuracy of the models was determined by mean square error (MSE), mean absolute deviation (MAD), bias and R Square. The datasets gathered from an experiment of Shre DTV at flume were divided into training data and testing data for the purpose of developing and validating all type of models. The difference between the model and the observed value become the forecasting error measurements. For the training data, the lowest MSE, RMSE and better R Square were noted for the Gompertz model. But ANN generalized better on testing data by obtaining lowest MSE, RMSE and higher R Square among others. ANN generalization result is 88.60%, Gompertz is 54.89%, curvilinear is 69.28% and static analytic is -1.29%. Lower bias was also for the neural network test data. As demonstrated by the bias values, only curvilinear model presenting overestimation model while other models produce little or no overestimation of the observed tilt response. Interpretations of the parameters estimation on Gompertz model have been attempted previously. However, focusing on the ability of Shre DTV to predict responses may be more practical than the relevance of parameter estimates. Universiti Teknikal Malaysia Melaka 2016 Article PeerReviewed Muharram, I. A. M. and Ismail, Z. H. (2016) Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method. Journal of Telecommunication, Electronic and Computer Engineering, 8 (11). pp. 1-6. ISSN 2180-1843 https://www.scopus.com/inward/record.uri?eid=2-s2.0-85011371139&partnerID=40&md5=687cc6f69e1d7081d11c85b73c7dd7d2 |
| spellingShingle | TK Electrical engineering. Electronics Nuclear engineering Muharram, I. A. M. Ismail, Z. H. Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method |
| title | Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method |
| title_full | Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method |
| title_fullStr | Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method |
| title_full_unstemmed | Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method |
| title_short | Modelling of shre drag tilt velocimeter (DTV) with curvilinear, gompertz and artificial neural network method |
| title_sort | modelling of shre drag tilt velocimeter dtv with curvilinear gompertz and artificial neural network method |
| topic | TK Electrical engineering. Electronics Nuclear engineering |
| work_keys_str_mv | AT muharramiam modellingofshredragtiltvelocimeterdtvwithcurvilineargompertzandartificialneuralnetworkmethod AT ismailzh modellingofshredragtiltvelocimeterdtvwithcurvilineargompertzandartificialneuralnetworkmethod |