sCARS-LSTM Model for Energy Consumption Prediction of Heat Setting Machines

The paper proposes an energy consumption prediction model based on sCARS-LSTM to solve the problem of difficult energy management of the heat setting machine in textile dyeing and finishing industry. Firstly, the sCARS screening method is used to screen the key factors that affect energy consumption, with the dimensions of the external input samples are compressed from 277 to 34, which is reduced by 87.7%. Then, the LSTM network is used to learn and predict energy consumption, and the prediction error RMSE of model used sCARS gets
$125.96\ \text{kW}\cdot \mathrm{h}$
, which is 9.36% lower than the best-performing baseline model. The proposed model has excellent predictive performance.