关键词: 洗衣机偏心, 数据驱动, 偏心预测, 低比特量化方法

Abstract: The smart home industry is developing rapidly,and the large-capacity and intelligent features have become the main development trend of drum washing machines. Operational failures caused by misjudgment of clothing imbalance are a core issue limiting the reliability. The key to solving this problem lies in accurate unbalance prediction. Load unbalance causes periodic drumvibration, variations in motor drive power, and fluctuations in motor speed. Currently, the mainstream method is the numerical lookup table method,whose disadvantage is that it relies on predefined models and fixed threshold values. It is difficult to adapt to the nonlinear unbalance characteristics induced by multiple factors such as clothing material and mass, making it suffer from insufficient prediction accuracy and robustness. This paver proposes a motor operating data-driven unbalance prediction method for drum washing machines. A dataset collection method for the model is proposed,and a solution is provided to eliminate the interference of experimental randomness and measurement noise on the quality of the dataset. A lightweight network architecture is designed and combined with a low-bit quantization method for deployment on low-cost control chips. The test platform is established. Experimental results demonstrate that the proposed prediction model achieves significantly higher accuracy than the current mainstream method,providing a reliable pathway for the stable operation of drum washing machines.

Key words: washing machine unbalance, data-driven, unbalance prediction, low-bit quantization method