Status : Verified
| Personal Name | Calimlim, Vincent A |
|---|---|
| Resource Title | Data Reconciliation using Hybrid PSO and Simplex with Gross Error Detection using Correntropy |
| Date Issued | 13 June 2024 |
| Abstract | A data reconciliation method using hybrid optimization method involving Particle Swarm Optimization together with Nelder-Mead Simplex Method is formulated. The proposed method is initially applied to a linear steady-state data reconciliation problem without gross errors and where all variables are measured, using a weighted least square error as objective function. Then it is extended to a problem where some variables are unmeasured. Then it is further applied to a system where some variables are unmeasured and some variables contains gross error. In this latter case, we know the gross error a priori. The proposed method is benchmarked with the results of Nelder-Mead Simplex method and PSO method. The results of the two methods, that of PSO and hybrid, approached global optimum. The results also show fewer runs for the proposed hybrid method to find global minimum than PSO method. For problems involving gross errors that are not known a priori, the objective function (estimator) needs to be modified. Fair, QWLS and correntropy estimator with the proposed hybrid method are evaluated in their ability to deal with effects of increasing gross error. All estimator were able to detect and identify the variable with gross error. The correntropy estimator showed best performance data reconciliation in terms of RER. |
| Degree Course | Master of Science in Chemical Engineering |
| Language | English |
| Keyword | data reconciliation; gross error detection; Nelder-Mead Simplex; Particle Swarm Optimization PSO |
| Material Type | Thesis/Dissertation |
Preliminary Pages
664.57 Kb
Category : F - Regular work, i.e., it has no patentable invention or creation, the author does not wish for personal publication, there is no confidential information.
Access Permission : Open Access