Strategic slot-pole optimization in electromagnetic coupling with random forest regressor model inspired by response surface methodology

This research introduces a novel approach for optimizing electromagnetic torque conversion in automotive powertrains by systematically evaluating pole-slot combinations and winding layouts. A hybrid data-driven strategy is employed: first, a random forest regressor models the complex, non-linear rel...

Full description

Saved in:
Bibliographic Details
Main Authors: Garbe Chukulu Jarso, Ramesh Babu Nallamothu, Rajendiran Gopal, Gang Gyoo Jin
Format: Article
Language:English
Published: Elsevier 2025-09-01
Series:Results in Engineering
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2590123025017116
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This research introduces a novel approach for optimizing electromagnetic torque conversion in automotive powertrains by systematically evaluating pole-slot combinations and winding layouts. A hybrid data-driven strategy is employed: first, a random forest regressor models the complex, non-linear relationships between key parameters (winding factor, symmetries, slots per pole per phase, balancing factor, and periodicity) with polynomial response surface fitting for analytical optimization. The RFR models achieved excellent predictive accuracy (R² up to 1.000 for deterministic variables), while the RSM enabled efficient gradient-based optimization of the design space. The optimized 8-speed configuration demonstrated precise gear ratio matching (<5 % deviation) while maintaining high electromagnetic performance (0.85–0.88 normalized response values). Analysis revealed critical design trade-offs: lower diametrical ratios (λ = √(1/3)) supported 83-pole designs for maximum torque density, while higher ratios (λ = √(1/1.25)) enabled more manufacturable ≤25-pole configurations. The hybrid ML approach, demonstrating the robust cross-validation performance (CV R² > 0.92). This work provides: (1) a novel framework combining data-driven modeling with classical optimization for electromechanical systems, and (2) practical guidelines for balancing performance and manufacturability in pole-slot design. The methodology's ability to efficiently explore high-dimensional design spaces while maintaining physical interpretability represents a significant advancement for developing next-generation vehicles and industrial power transmission systems.
ISSN:2590-1230