Robust design optimization of spin algorithm to reduce spinning time and vibration in washing machine

• Published in: AIP advances 2024-03, Vol.14 (3), p.035219-035219-11
• Main Authors: In, Daekil, Lim, Dae-Guen, Min, Junhong, Kim, Seung-Wook, Seo, Hyunseok, Kim, Sungmin, Choi, Dong-Hoon, Cheong, Cheolung
• Format: Article
• Language: English
• Subjects: Algorithms; Computer simulation; Design optimization; Monte Carlo simulation; Optimization algorithms; Performance measurement; Random variables; Reduction; Robust design; Simulation models; Uncertainty; Vibration; Washing machines
• ISSN: 2158-3226; 2158-3226
• DOI: 10.1063/5.0188132

This study focuses on optimizing the spin algorithm of washing machines to reduce spin time and vibration, two crucial performance metrics often in a trade-off relationship due to structural limitations. We developed a simulator for the spin process that accounts for the uncertainty associated with the clothes’ position in the drum. Using this simulator, we designed a robust optimization algorithm to minimize spin time while maintaining acceptable vibration levels. The variation in the mass and position of laundry within the drum primarily affects vibration performance during the spin process. We formulated the unknown unbalance caused by the laundry as test-implementable unbalance parameters (UBP), represented as random variables. The uncertainties in the laundry’s unbalance for repeated experiments were estimated by the distribution of these UBPs. Experiments were conducted at parametrically designed points to measure vibrations and internal sensor values for control in specific revolutions per minute (RPM)ranges. These data were used to create approximate models. Monte Carlo simulation was then applied with UBP distributions obtained from the clothing experiments to meta-models, and the simulator was equipped with the spin algorithm. This process allowed us to predict the distribution of the maximum tub vibration displacement and spin time. A robust optimization algorithm was subsequently applied to the simulation model to derive the optimal spin algorithm. The proposed methodology yielded substantial improvements: a 25% reduction in the mean value of maximum displacement and a 90% reduction in the mean value of spin time compared to initial values.