Whale optimization algorithm based on tent chaotic map for feature selection in soft sensors
Irrelevant features in data collected from oil refineries affect system performance due to conflicts between normal data and detected faults. Selecting the relevant features from the data leads to better classification results. Optimization algorithms are successfully applied in the feature selectio...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Dr D. Pylarinos
2025
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/121032/ http://psasir.upm.edu.my/id/eprint/121032/1/121032.pdf |
| Summary: | Irrelevant features in data collected from oil refineries affect system performance due to conflicts between normal data and detected faults. Selecting the relevant features from the data leads to better classification results. Optimization algorithms are successfully applied in the feature selection task in many systems. One of the powerful optimization algorithms that is used for feature selection is the Whale Optimization Algorithm (WOA), which is a nature-inspired metaheuristic optimization algorithm that mimics the social behavior of humpback whales. This study presents an improvement to WOA using a tent chaotic map to select the most relevant features and enhance performance. The Tent map mainly applies randomness to generate diversification into the search process and escape from local optima in WOA. The tent map is used for generating the initial population, producing values in control parameters, and updating the position of search agents. The proposed approach combines the tent map with WOA, called TWOA, to enrich population diversity, prevent premature convergence, and speed up convergence. TWOA is applied in a soft sensor with actual data collected from the Salahuddin oil refinery in Iraq. The soft sensor was designed using several stages, including data collection, preprocessing, clustering, feature selection, and classification. The proposed TWOA achieved a higher fault classification result of 99.98% compared to other algorithms. |
|---|