Unbalanced Load Estimation Algorithm Using Multiple Mechanical Measurements for Horizontal Washing Machines

In horizontal washing machines, accurate estimation of unbalanced loads, including their magnitudes and locations in washing drums, is always required to achieve quiet highspeed spinning operations. Many estimation algorithms have been developed for the purpose and their estimation accuracy needs to be evaluated in real laundry processes. Nevertheless, lack of true information on the unbalanced loads during the process against which different algorithms could compare makes the evaluation a nontrivial task: it is very hard, if it is not impossible, to directly quantify the actual unbalanced load magnitude/location while the laundry is being washed. To tackle the problem, we use multiple mechanical measurements from laser displacement sensors, accelerometers and load cells which are installed in different locations on a horizontal washing machine, to derive true unbalanced load information inside a drum. Neural networks with PCA preprocessing are employed for the estimation and experiments have been designed to train and test the neural networks. Performances of the neural network-based estimation algorithm are reported.